Based on #866

This uses async-executor under the hood and is highly inspired by both smol and the async-executor implementation inside of async-std.

Pros:

• less code inside async-std, simpler integration
• shares the same executors with any other users of async-global-executor
• allows spawning local tasks on all of the global executor threads

Cons:

• adds one more extra dependency
• ???

Hi, thanks for this PR!

I def see how sharing the global executor between projects could be useful. But in order to accept this PR I'd like to see:

1. async-global-executor dual licensed as MIT / Apache-2.0. If ever need to migrate away from async-global-executor we shouldn't need to worry about licensing questions.
2. Granting commit/publish access to @dignifiedquire, @stjepang or myself. Or moving the project to the async-rs org. That way if something ever needs fixing that is time-critical and for some reason we can't reach you, we're still able to move ahead.

I hope that makes sense. Thanks heaps!

Done and Done.

I'm fine with moving it to async-rs too if you see fit

Rebased on top of the 1.0 bumps

Thanks. I was out for the last two weeks, will be back in the next days and take a look at this.

Rebased

CI failure should now be fixed with async-executor 1.1.1

Looks fine to me, given that this is pretty much still the same code, but in a different place for now. I do like having it extracted as its own crate.
Thanks