perf: process: replace timeout with queueMicrotask

[ThaUnknown]

This old process polyfill uses a fix for setTimeouts which was originally made to work on some insanely niche smart TV's. We don't care about that as our minimal entry point is ESM anyways, so I replaced it with queueMicrotask which has very similar support ranges. This was first done by mathiasvr in process-fast so I changed the copyright to his, even tho it's just 1 line of code.

Why this matters?
I'll quote Endless on this matter:
The problem is with process browserified version of process.nextTick it uses a setTimeout(fn, 0) as a fallback. The setTimeout is delayed well enough in foreground to strangle the speed from 6-7000 kb/s down to just 400 kb/s when one tab is hidden "to save resources". if you instead use two windows the speed will be fast (normal) again.

setTimeout(fn, 0) can potentially be used, however as it is clamped to 4ms for timers nested more than 5 deep per the HTML spec, it does not make for a suitable polyfill for the natural immediacy of setImmediate. - MDN

- link

CPU usage aside, unlike setTimeout, queueMicrotask takes ~ microsecond to execute, so in those worst case scenarios it would be over x10000 faster than setTimeout.

Optionally if backwards compatibility is an issue, an absolutely tiny queueMicrotask polyfill could be used

[ije]

LGTM, thanks!

[ThaUnknown]

I'm not certain why tests fail, demo has queueMicrotask so I don't see the issue