fix(maitake-sync): add proper ScopedRawMutex bound to wait_map::Wait

[peterkrull]

I was unable to await the Wait when using a mutex defined in the mutex crate because a trait bound was not met. The current trait impl is not general enough.

[peterkrull]

This seems to also be a thing for other types which specifically use the DefaultMutex default type, like the Lock for the async mutex.

pub struct Lock<'a, T: ?Sized, L: ScopedRawMutex = DefaultMutex> { .. }

impl<'a, T> Future for Lock<'a, T> {..}

[hawkw]

Whoops! Thanks for fixing this!

We should probably add some tests that actually construct each synchronization primitive with an overridden ScopedRawMutex type and check that all the expected trait impls are present, to prevent against future regressions with this sort of thing. I don't think it's worth holding up merging this fix for such tests, but I'll open an issue because I think it's worth doing.

[hawkw]

Opened #516 for adding tests to ensure expected Future impls are present.

[hawkw]

This seems to also be a thing for other types which specifically use the DefaultMutex default type, like the Lock for the async mutex.

pub struct Lock<'a, T: ?Sized, L: ScopedRawMutex = DefaultMutex> { .. }

impl<'a, T> Future for Lock<'a, T> {..}

#517 fixes this for the mutex::Lock future. Checking the other async synchronization primitives (WaitQueue, RwLock, and Semaphore), it looks like their Future types already have proper trait bounds, but I'll add regression tests for them as well.