Does it make any sense for trio to support micropython?

I don't really know anything about micropython, but @pfalcon recently brought to my attention that they have their own incompatible variant of asyncio, uasyncio, which makes interesting choices like leaving out Futures altogether.

This leads me to wonder: how well trio would work on micropython? I'm not thinking of any ridiculously spendthrift data structures... we do currently splurge on a sortedcontainers.SortedDict for tracking timeouts instead of a traditional heap, in order to get O(1) removal of stale entries, but that's a pretty self-contained bit of code and certainly not visible to users. OTOH there are probably another thousand things I'm not thinking of that would make microcontroller devs gasp in horror :-). Anyway, it'd be interesting to know what would be required to make it work, or why concretely it couldn't work, so maybe we can collect any notes on that here in this issue.

No files, tests, or entry points are identified. Start by comparing Trio's stated dependencies and implementation assumptions with MicroPython and its uasyncio variant, including the SortedDict timeout-tracking concern; done means documenting concrete compatibility requirements or blockers.