Allow async-await macros to be used without std

[arcnmx]

select!() is the only macro that requires std due to implementation details, so allow everything else through.

[arcnmx]

I wanted to also address random on no-std but it seemed a bit too intrusive of a change for me to make all the design decisions...

• armv6m only has AtomicUsize::{load, store} which means reducing the xorshift state width depending on target pointer size
• Also means being okay with the state potentially repeating in the cases where calls overlap during the load+store and clobber each other?
• I think I actually want a non-randomized select!() anyway that polls in priority order so would need to reimplement select anyway.

I'm probably reading too much into it given how weak the randomness requirement here really is but I left that out for now... though I am tempted to leave the macro itself in so it could still be used if you override the random fn using a facade.

[taiki-e]

(FYI: There is a discussion in #1805 about adding a select that doesn't use randomness. and seems implementation is not difficult: taiki-e@2691aa2)

[arcnmx]

Yeah that would be really nice, is there anything blocking that from becoming a PR? just docs? or name, default, syntax bikeshed? uncertainty around whether it should be provided?

[taiki-e]

is there anything blocking that from becoming a PR? just docs? or name, default, syntax bikeshed? uncertainty around whether it should be provided?

I didn't open the PR because the document was incomplete, but in practice, we will need to decide which is the default.

[arcnmx]

Yeah, I can see the argument made for keeping unbiased as the default, though don't have any personal opinions there as long as the ability to choose can be made available.

Also, I threw up what I had for no-std random and could PR in case it's something that could be accepted?

[martell]

@arcnmx firstly thanks for looking at this,
Is there anything stopping this from being merged?
I'm looking forward to using async await without std upstream.

[arcnmx]

Rebased onto master, so select_biased! is now an option on no-std with these changes. select! still requires #1912 but the biased alternative being available makes it a bit less of a blocker now.

[arcnmx]

@taiki-e you've approved this previously, any chance it could be considered for merging? I've been experimentally making use of these changes for quite a while now and it'd be nice to have these available upstream - it seems like a benign change, and select_biased has made it even more appealing now.

(cc @Nemo157 as this is somewhat your domain as well)

[taiki-e]

I'd like to merge this. cc @cramertj

[cramertj]

This looks fine to me, thanks for the ping! I'll get a release out shortly.