fix(eval): strip query strings when extracting Wikipedia titles from URLs

[longzhenren]

Root cause

WikipediaAPIRetriever._extract_wiki_title strips only the #fragment suffix from Wikipedia URLs:

pattern = r"https?://[a-z]{2,3}\.wikipedia\.org/wiki/(.+?)(?:#.*)?$"

Mobile share links almost always carry query parameters (e.g. ?wprov=rarw1), so the extracted title kept the query string and the downstream retrieve() call looked up a non-existent article, degrading retrieval quality.

Repro on main:

url = "https://en.wikipedia.org/wiki/Albert_Einstein?wprov=rarw1"
# before: "Albert Einstein?wprov=rarw1"
# after:  "Albert Einstein"

Fix

Strip both query and fragment suffixes: the pattern tail becomes (?:[?#].*)?$.

Test

Added tests/test_wiki_title_extraction.py covering: plain URLs, query-string stripping, fragment stripping, percent-encoded titles, and non-Wikipedia URLs returning None.

[vercel]

@longzhenren is attempting to deploy a commit to the andylizf's projects Team on Vercel.

A member of the Team first needs to authorize it.