Refactor the analog module

[jessebraham]

This is still largely untested, so opening as a draft for now, but will get it verified early next week. Just wanted to get some eyes on it in the meantime.

This refactoring had a few goals: improved code organization, reduced duplication, and improved documentation, primarily.

There are quite a few changes, but they are mostly pretty straight-forward.

Regarding the analog::dac module, I was able to eliminate the macro entirely and shrink the line count of the module overall, so that's a big win IMO. This came at the cost of a small amount of duplication, however it's quite minimal and this module is not likely to change significantly any time soon, so I think it's a reasonable tradeoff.

The analog::adc module was re-organized a bit, and I was able to extract out some common types from the various implementations, further reducing duplication. I also updated the documentation, which looks reasonably good now and has examples for both peripherals. I've additionally decoupled this driver from the embedded-hal traits while I was in here, by introducing the AdcChannel trait.

The peripheral constructors have changed a bit, and you no longer need to split a peripheral to get access to the ADCs/DACs, but other than that there aren't really any user-facing changes here, I don't think.

[jessebraham]

Did some really quick testing and things at least seem to still be working 😁

[bjoernQ]

I think I like the idea to get rid of the .split() - I wonder if it would be a good idea to do the same with SYSTEM.split()?

Overall looks good to me - changes things to the better. Maybe just consider hiding / sealing internal helper traits

[JurajSadel]

LGTM, thanks!