allow reversing the area chart fill baseline

[samfrjn11]

Currently, flux:chart.area's fill always renders from the line down to the axis minimum (domain[0]), with no way to configure it otherwise. This works well for typical "value grows from
zero" charts, but breaks down for data where larger magnitude means worse, not better — the most common case being RF/network signal strength in dBm, which is always negative (e.g. -30 dBm
= excellent, -130 dBm = no signal).

For that kind of data, the natural chart convention is to put the best reading at the top of the y-axis and the worst at the bottom (which Flux already does correctly, since it sorts
numerically and -30 > -130). But because the area fill baseline is hardcoded to whichever value sits at the bottom of the axis, the fill ends up anchored to the worst reading — meaning the
shaded area is large during good signal and shrinks to nothing during bad signal. For a chart whose whole purpose is to highlight how bad a signal got, that's backwards.

Workaround we're using today

Plot abs(value) instead of the real signed value (so the axis's numeric ordering flips), and use tick-prefix="-" to restore the correct-looking negative labels. It works, but it's a hack —
it also means we now have to pin tick-start/tick-end explicitly, since the "auto-anchor to 0" behavior (tickStart === "auto" && min > 0 && max > 0) kicks in for all-positive plotted
values and produces a meaningless "0" tick.

Requested feature

A way to control the area's fill baseline independently of which end of the domain is "first" — e.g.:

• A baseline prop on flux:chart.area: baseline="min" (current/default), baseline="max", or an explicit baseline="{value}"
• Or a reverse option on flux:chart.axis that flips which domain endpoint the area treats as its baseline, without changing tick/label ordering

Either would let us drop the abs-value workaround entirely.

Desired result: