Audit timers HAL #1: Refactor timer wakers

[kurtjd]

Splitting this up into two or three PRs just to make reviewing easier. This first part refactors wakers so we no longer need to index into an array and instead each waker is associated with each peripheral via the info method like other peripherals.

The interrupt handler will instead call the binded peripheral's T::info().wake(...) method with the timer type and channel that generated an interrupt just to be certain that we actually want to wake. This has the benefit of simplifying some of the code as well.

Tested via rt685s-evk/timer.rs example on hardware (tested counting timer and capturer time by connecting signal to eval board). The blocking wait_us:

embassy-imxrt/examples/rt685s-evk/src/bin/timer.rs

Line 39 in a5f5813

tmr1.wait_us(3000000); // 3 seoconds wait

doesn't seem to work (it blocks indefinitely) but it also doesn't work just using tip of main, so seems like it's always been broken?

Related: #530

[jerrysxie]

@kurtjd Yeah, blocking counting down timer is broken due to it being reliant on interrupt. After the interrupt refactoring, it is broken. Another way to do without relying on interrupts would be to set MR0R and MR0S to enable reset + stop on match and then check for TC being reset to 0.

[kurtjd]

@kurtjd Yeah, blocking counting down timer is broken due to it being reliant on interrupt. After the interrupt refactoring, it is broken. Another way to do without relying on interrupts would be to set MR0R and MR0S to enable reset + stop on match and then check for TC being reset to 0.

To clarify, do you mean there was a previous interrupt refactoring that broke it (before this PR) and it never got fixed? Or that my refactoring broke it? It seemed to not work even from main so I didn't think my changes broke it, but can double check.

Edit: Nevermind, found the background context here: #486. Fixing this would probably be best for a separate PR then.

[williampMSFT]