zephyr:embassy: Use a semaphore for the executor

[aolowin]

Fixes a potential race condition in the suspend/resume sequence.

[d3zd3z]

@aolowin I've approve the workflow to run, so you should see the rustfmt fixes that will be needed.

Also, you will need a Signed-off-by line in the commit, which you should be able to add with git commit --amend -s.

The other minor nit is that there should be a space after the colon in the commit title, e.g. zephyr: embassy: ....

[aolowin]

I'm not sure why the build is failing on Semaphore::new(0, 1).unwrap()

The unwrap() is needed on my build to compile and it's used in a few of the samples.

[mrodgers-witekio]

I'm not sure why the build is failing on Semaphore::new(0, 1).unwrap()

I think this commit removed the need for unwrap: 3a0016d

[aolowin]

I'm not sure why the build is failing on Semaphore::new(0, 1).unwrap()

I think this commit removed the need for unwrap: 3a0016d

Indeed. My build directory was a few commits behind. Thanks!

[d3zd3z]

My testing shows this to work well. I like the solution, and this will make it easier to add support for async wait on other semaphores.

[cfriedt]

LGTM

[d3zd3z]

I'm not sure why the build is failing on Semaphore::new(0, 1).unwrap()

The unwrap() is needed on my build to compile and it's used in a few of the samples.

This was unfortunately needed to allow semaphores to be declared as statics, which needs the construct to be const. Result is technically returnable from const, but it is almost impossible to use (at least with stable rust). So, I changed the behavior to panic if the initializer is nonsensical, and just return the Semaphore.