Fix RealTimeClock & UsbBus ownership

[jnthbdn]

The RealTimeClock works great, but it takes ownership of the rtc_clock field in the ClockManager, which makes it invalid (partial move).
Moreover, the RealTimeClock struct doesn't allow freeing rtc_clock, the new function doesn't do anything about it, so the instance is destroyed.

Following the model of other functions present in the HAL for clock configuration (such as setup_pll_blocking or configure_clock from ClockManger), I modified the parameter so that it was a reference.

#724

Thanks for this brilliant HAL!

[ithinuel]

Would it make sense to make RealTimeClock only take a frequency here instead?

A benefit of expecting the RtcClock is that at least it implicitly checks one's using the right clock source; but it doesn't tie them together either.
Nothing prevent that clock source to be reconfigured which would render anything the RealTimeClock inconsistent.
So since the benefit provided by using &RtcClock is pretty thin, is it worth keeping at all?

cc @thejpster @9names @jannic

[jnthbdn]

Thanks for your review, I quite agree with your comment, it would make the function's intentions clearer.

But the more I look at the code, the more I wonder if my PR is really relevant...

Let me explain, as you said in issue #724, it's interesting that ownership is taken in order to ensure that clocks are not subsequently modified. But there's still the problem of the ClockManager being partially moved. So in the end, even though it's a big change, wouldn't it be better to make the ClockManager fields Option<T>?
Changing this line

rp-hal/rp2040-hal/src/clocks/macros.rs

Line 22 in cfd6c12

pub [<$name:snake>]: $name,

with

pub [<$name:snake>]: Option<$name>

Because in this PR we've solved the RealTimeClock problem, but the new function in the UsbBus does the same thing.

rp-hal/rp2040-hal/src/usb.rs

Line 441 in cfd6c12

_pll: UsbClock,

[ithinuel]

Would it make sense to rather than having fields public, adding a split method and include as part of its output a extra entry offering a subset of what the clock manager does but that does not collide with the individual clocks ?

[jnthbdn]

I modified the PR by adding free functions to the RealTimeClock and UsbBus structs.

[ithinuel]

LGTM