Expose sleep-wait functions.

[davidchisnall]

We didn't have generic functions for spinning / sleeping and they're quite important.

[nwf-msr]

These are fine, I guess, even if I generally think large waits should be CPS hung off timers and short waits shouldn't be a thing. Where did you find you needed them?

[davidchisnall]

These are fine, I guess, even if I generally think large waits should be CPS hung off timers and short waits shouldn't be a thing. Where did you find you needed them?

Short sleeps are normally needed (in interrupts-disabled things) for bit banging. The longer one is needed when initialising the Ethernet PHY (requires a 167ms sleep, which can be a wait for a few ticks and then spin for any remaining time).

[davidchisnall]

I generally think large waits should be CPS hung off timers

This is also necessary for implementing a timing wheel for CPS callbacks. We may want to separate out the sleep-delay calculation so that it can be integrated with a timeout for event-driven code.