Add standalone waker constructor for executor and expose it

[Ddystopia]

Currently we can spawn a task with task::spawn(), but there is no way to get a waker. This PR adds const function task::waker(), that returns a waker that will wake that task.

My use case for this is smoltcp, they expose a lot of functions like that one: Socket::register_waker

[korken89]

Hi! Thanks for the PR.

Can you motivate the need for this a bit more? As I see it, and e.g. what embassy-net does internally (https://github.com/embassy-rs/embassy/blob/main/embassy-net/src/lib.rs#L797) is the following to not need to have a way to get a waker:

async fn do_something() {
    poll_fn(|cx| {
        smoltcp.register_waker(cx.waker());

        if smoltcp.api_that_later_will_wake().is_done_or_something() {
            Poll::Ready(())
        } else {
            Poll::Pending
        }
    }).await
}

[Ddystopia]

Hi! Thanks for the PR.

Can you motivate the need for this a bit more? As I see it, and e.g. what embassy-net does internally (https://github.com/embassy-rs/embassy/blob/main/embassy-net/src/lib.rs#L797) is the following to not need to have a way to get a waker:

async fn do_something() {
    poll_fn(|cx| {
        smoltcp.register_waker(cx.waker());

        if smoltcp.api_that_later_will_wake().is_done_or_something() {
            Poll::Ready(())
        } else {
            Poll::Pending
        }
    }).await
}

Sometimes I need to wake other task than the one setting the waker, or with some other logic. So I would need to create the waker with poll_fn and store it somewhere, then retrieve with other methods. And it can't be done in const contexts. With that change code is very straightforward by directly using the waker for a task.

So yes, it is possible, for non-const use cases, to use poll_fn to extract the waker, but when use case is more complicated then a basic one, user of the api would suffer, while it is really trivial for rtic to add that getter, as presented in pr.

[Ddystopia]

@korken89 is this motivation enough? This allows to not require a poll and not store a waker somewhere if you are not using it directly inside, as embassy-net is doing. This PR is really small while ergomic gains are larger than PR.

[korken89]

It's no big addition, I'll add it. 👍

[korken89]

@Ddystopia can you fix the conflicts and I'll merge?

[AfoHT]

Rebased and moved changelog entries to Unreleased (releases happened) :)

Thanks!