copy(garage): cut the staging tics declauding 0.13.0 flags

src/pages/garage/blueprint.html

@@ -288,7 +288,7 @@ <h2>What I would not touch<span class="sev keep">the soul</span></h2>
     </ul>
 
     <div class="callout">
-      <strong>The verdict, honestly.</strong> Half of this blueprint is already the changelog:
+      <strong>The verdict.</strong> Half of this blueprint is already the changelog:
       the Workers migration, the in-house counter DO, granular routing, the build step, the
       edge-cache kit all landed incrementally, each gated by a 42-route oracle. The site
       never went down for any of it. That is the actual argument against a big-bang rewrite:

src/pages/garage/chunks.html

@@ -127,7 +127,7 @@ <h3>FastCDC <small>content-defined: min 16 · avg 40 · max 120 bytes</small></h
 
     <h2>2 · FastCDC, briefly</h2>
     <p class="note">
-      Here is the whole idea, with every operational concern taken back out:
+      Here is FastCDC with every operational concern taken back out:
     </p>
     <pre class="toy">function cut(bytes, avg) {
   var mask = avg - 1, hash = 0, cuts = [];
@@ -157,7 +157,7 @@ <h2>2 · FastCDC, briefly</h2>
 
     <h2>3 · Big files</h2>
     <p class="note">
-      Content-addressing earns its keep on large blobs. Drop any file (it
+      Content-addressing pays off on large blobs. Drop any file (it
       never leaves your machine) and your browser FastCDC-chunks it locally
       (avg 8&nbsp;KB), then <a class="term" data-t="sha-256" href="https://en.wikipedia.org/wiki/SHA-2" title="A hash function that turns any input into a 256-bit fingerprint. Change one byte of the input and the fingerprint changes completely." rel="external">SHA-256</a>'s each chunk into a content ID. The blocks
       run proportional to size so you can see the variable-length cuts; the

src/pages/garage/cloudflare.html

@@ -53,13 +53,13 @@
   <main id="main" class="content prose">
     <h1>Four free Cloudflare features, live</h1>
     <p class="lede">
-      The free Workers plan quietly bundles things that used to cost money or
+      The free Workers plan bundles things that used to cost money or
       a separate service. I wired four of them to live demos here
       (<b><a class="term" data-t="durable-object" href="https://developers.cloudflare.com/durable-objects/" title="A Cloudflare Workers object with its own built-in storage, where each named instance lives in one place so every request to it is coordinated." rel="external">Durable Objects</a></b>, <b><a class="term" data-t="workers-ai" href="https://developers.cloudflare.com/workers-ai/" title="Cloudflare's service for running machine-learning models, such as language and vision models, on its own GPUs, called directly from a Worker." rel="external">Workers&nbsp;AI</a></b>, <b><a class="term" data-t="workers-logs" href="https://developers.cloudflare.com/workers/observability/logs/workers-logs/" title="Cloudflare's built-in log store for Workers. It collects each request's log lines so you can search and filter them in the dashboard." rel="external">Workers&nbsp;Logs</a></b>, and <b><a class="term" data-t="trace-span" href="https://opentelemetry.io/docs/concepts/signals/traces/" title="One timed step inside a traced request, with a name, a start time, a duration and attributes. Nested spans show where the time went." rel="external">custom trace spans</a></b>),
       behind a tiny worker (<code class="m">cf-garage</code>) routed at
       <code class="m">/garage/cf/*</code>. The buttons hit real endpoints; nothing is faked.
       A fifth, <b><a class="term" data-t="browser-run" href="https://developers.cloudflare.com/browser-run/" title="Cloudflare's service for controlling a headless browser from a Worker, to render pages, take screenshots or make PDFs." rel="external">Browser&nbsp;Rendering</a></b>, I bound and deployed, then parked on the free
-      tier&apos;s daily cap. <a href="#parked">Noted honestly below.</a>
+      tier&apos;s daily cap. <a href="#parked">The note below says why.</a>
     </p>
 
     <div class="cf-feat">
@@ -124,7 +124,7 @@ <h2>The honest scorecard</h2>
       request shows its own spans in the trace waterfall.</p>
 
     <div class="callout" id="parked">
-      <b>The fifth, parked honestly: Browser Run.</b> <a class="term" data-t="headless-browser" href="https://en.wikipedia.org/wiki/Headless_browser" title="A web browser that runs with no window on screen, driven by code, so a program can load pages, run their scripts and capture the result." rel="external">Headless Chromium</a> in a Worker
+      <b>The fifth, parked: Browser Run.</b> <a class="term" data-t="headless-browser" href="https://en.wikipedia.org/wiki/Headless_browser" title="A web browser that runs with no window on screen, driven by code, so a program can load pages, run their scripts and capture the result." rel="external">Headless Chromium</a> in a Worker
       via <code class="m">env.BROWSER.quickAction("screenshot", options)</code>, the basis for
       on-the-fly <code class="m">og:image</code> cards. I bind
       it and the worker deploys; no enable toggle exists (the binding

src/pages/garage/compression.html

@@ -180,7 +180,7 @@ <h2>Deltas, and why they ship as dcz</h2>
       because decode scales with the CPU while 44 bytes stays 44 bytes.
     </p>
     <p>
-      Worth saying plainly: on the old numbers this was a coin flip, and on these it's lopsided. The
+      On the old numbers this was a coin flip, and on these it's lopsided. The
       size table now favours brotli by more than it ever did, and that table is the wrong axis to
       decide on.
     </p>

src/pages/garage/encoding.html

@@ -335,14 +335,14 @@ <h3>Chrome paints nothing until every channel has started <span class="badge rev
     <p class="note"><b>On a thumbnail that barely matters.</b> The JPG tile is the fallback most
       browsers never fetch, and it is small: a median 48.7&nbsp;KB, with the gate at a median
       19.3&nbsp;KB. A 10-packet initial <a class="term" data-t="congestion-window" href="https://en.wikipedia.org/wiki/TCP_congestion_control#Congestion_window" title="How much data a TCP sender may have in flight before it hears back. Connections usually start at 10 packets and grow as replies arrive." rel="external">congestion window</a> carries about 14&nbsp;KB, so 116 of the 255
-      reach their gate in the opening flight and the rest one round trip later. The scan search still earns its place
+      reach their gate in the opening flight and the rest one round trip later. The scan search is still worth running
       at this size, purely for the bytes it saves.</p>
 
     <p class="note">It matters at <a href="/photos">the full-size archive</a>. <code>/images/full/</code>
       is the one image surface here with no AVIF tier and no <code>&lt;picture&gt;</code> around it:
       the worker streams a median 21.6&nbsp;MB progressive JPEG from R2 (since October 2026 as its
       lossless JPEG XL copy, about 8% smaller), and that single file is the whole payload. That is also the awkward part of the format story, because <b>AVIF
-      structurally cannot do this</b>. Its layered mode caps at four stages, paints early only in
+      can&apos;t do this</b>. Its layered mode caps at four stages, paints early only in
       Chromium, and is staged rather than incremental the way a JPEG scan script is. On the tiles it
       costs 6 to 10% more bytes (<a href="#layered-avif">measured below</a>), so nothing here emits
       it. So the format that wins every byte comparison on this page
@@ -406,7 +406,7 @@ <h3>What the archive tier actually costs</h3>
       Re-encoding at 4:4:4 instead of 4:2:2 scores 88.28 against 88.15, so the chroma round-trip is
       not the limit and 4:2:2 keeps its ~13% byte saving for free. Flattening the reference to 8-bit
       first scores 88.08 against 88.15, so the 10-bit&rarr;8-bit step is not the limit either: JPEG is
-      8-bit, and a HIF&apos;s extra depth earns its keep in the edit.</p>
+      8-bit, and a HIF&apos;s extra depth pays off in the edit.</p>
     <p class="note">That generation also costs far less than it looks once an image is resized for
       anywhere else. Comparing a direct downscale of the HIF against a downscale of the archive JPEG,
       the same pair that reads 88.15 at full resolution reads <b>94.62 at 1080px</b>: averaging ~50

src/pages/garage/hidden-flags.html

@@ -216,7 +216,7 @@ <h2>The inventory</h2>
 <h2>Two questions, and only one of them is free</h2>
 <p>When you find an undocumented option, there are two separate things to establish, and reading the docs answers neither.</p>
 <p><b>Is it accepted?</b> Does the tool recognise the key at all, or is it dropping an unknown name on the floor? <b>Does it work?</b> Does accepting the key change the behaviour the flag claims to change.</p>
-<p>The first question is usually free, and that is the part people skip. A tool with strict argument parsing answers it through its exit code alone, with no work performed and nothing deployed:</p>
+<p>The first question is usually free, and people skip it. A tool with strict argument parsing answers it through its exit code alone, with no work performed and nothing deployed:</p>
 <pre class="hf-code"><code>$ wrangler deploy --x-bogus-flag ; echo $?
 1
 $ wrangler deploy --x-provision=false --dry-run ; echo $?
@@ -230,8 +230,8 @@ <h2>When silence is the failure mode</h2>
 <h3>A payload schema that refuses unknown keys</h3>
 <p>Cloudflare's <a class="term" data-t="browser-run" href="https://developers.cloudflare.com/browser-run/" title="Cloudflare's service for controlling a headless browser from a Worker, to render pages, take screenshots or make PDFs." rel="external">Browser Run</a> binding takes a JSON payload. A launch post described selecting a rendering engine with one extra key; the Quick Actions reference did not list it. Sending it answered:</p>
 <pre class="hf-code"><code>{"code":"unrecognized_keys","keys":["browser"]}</code></pre>
-<p>An invented engine name returned the byte-identical error, which is the useful half of the result. The schema is closed, so it refuses the option rather than failing on the value, and a closed schema is a gift: it answers "is this accepted" for free, every time, with no render performed and no browser minutes spent.</p>
-<p>The same feature then produced the opposite trap through the <a class="term" data-t="rest-api" href="https://en.wikipedia.org/wiki/REST" title="A common style of web API where each URL names a resource and standard HTTP methods such as GET and POST act on it." rel="external">REST</a> door, where the parameter is merely ignored. A 200 came back either way, so the code labelled the render with the engine it had asked for. On a page whose entire premise is showing what a machine actually saw, that label was a claim nobody had checked. It reads <code>kitesurf-requested</code> now.</p>
+<p>An invented engine name returned the byte-identical error. The schema is closed, so it refuses the option rather than failing on the value, and a closed schema is a gift: it answers "is this accepted" for free, every time, with no render performed and no browser minutes spent.</p>
+<p>The same feature then produced the opposite trap through the <a class="term" data-t="rest-api" href="https://en.wikipedia.org/wiki/REST" title="A common style of web API where each URL names a resource and standard HTTP methods such as GET and POST act on it." rel="external">REST</a> door, where the parameter is merely ignored. A 200 came back either way, so the code labelled the render with the engine it had asked for. On a page whose premise is showing what a machine actually saw, that label was a claim the code never checked. It reads <code>kitesurf-requested</code> now.</p>
 
 <h3>A compression option three runtimes disagree about</h3>
 <p><a class="term" data-t="compression-dictionaries" href="https://www.rfc-editor.org/rfc/rfc9842.html" title="A scheme where the browser keeps an earlier copy of a file and the server sends only a compressed difference against it." rel="external">Dictionary compression</a> takes a <code>dictionary</code> option. <a class="term" data-t="nodejs" href="https://en.wikipedia.org/wiki/Node.js" title="A program that runs JavaScript outside the browser, widely used for servers and build tools." rel="external">Node 26</a> honours it. <a class="term" data-t="cloudflare-workers" href="https://developers.cloudflare.com/workers/" title="Code that runs on Cloudflare's servers around the world and answers each request directly, with no server of the site's own." rel="external">Workers</a> ignore it for <a class="term" data-t="zstd" href="https://en.wikipedia.org/wiki/Zstd" title="Zstandard, a compression algorithm from Meta built for very fast decompression. Browsers accept it as a way to shrink web pages in transit." rel="external">zstd</a> and say nothing. <a class="term" data-t="bun" href="https://en.wikipedia.org/wiki/Bun_(software)" title="A JavaScript runtime that also bundles a package manager, bundler and test runner, built to work as a replacement for Node.js." rel="external">Bun</a> ignored it through 1.3.14 and fixed it in <a href="https://github.com/oven-sh/bun/pull/34427" rel="external">oven-sh/bun#34427</a>. Every one of those failures is silent, because a frame compressed <em>without</em> a dictionary still decodes perfectly <em>with</em> one. Nothing throws. The only symptom is a byte count that never got smaller.</p>
@@ -394,7 +394,7 @@ <h2>The rule this leaves</h2>
 <p>And prefer reading the definition to running the experiment when the definition is in your <code>node_modules</code>. Two of the four cases here were settled by a source comment and a call site, in less time than a deploy takes, and neither answer could have been reached by watching the tool succeed.</p>
 
 <h2>What would change this</h2>
-<p>A tool that accepts unknown flags silently would break the free half outright, and the check for that is the bogus-flag control itself: an exit code of 0 on an invented name means every conclusion on this page has to be re-earned with real runs. A hidden flag becoming documented would move its row from this list to an ordinary reference, which is the outcome to hope for. And a measurement showing the capture does honour <code>waitForTimeout</code> would close the one case here that is deliberately still open.</p>
+<p>A tool that accepts unknown flags silently would break the free half, and the check for that is the bogus-flag control itself: an exit code of 0 on an invented name means every conclusion on this page has to be re-earned with real runs. A hidden flag becoming documented would move its row from this list to an ordinary reference, which is the outcome to hope for. And a measurement showing the capture does honour <code>waitForTimeout</code> would close the one case here that is deliberately still open.</p>
     <section id="luq" aria-label="understanding check"></section>
   </main>
   <div class="xp-statusbar"><span>garage</span><span>shipped</span><span>2026-08-24</span></div>
@@ -442,7 +442,7 @@ <h2>What would change this</h2>
       ]
     },
     {
-      "q": "The dictionary control runs three times: no option, the right value, a wrong value. Why does the wrong value earn its place?",
+      "q": "The dictionary control runs three times: no option, the right value, a wrong value. Why run the wrong value at all?",
       "options": [
         {
           "t": "It ties the smaller output to the value you passed rather than to some other cause.",