Unstabilize CPU-related functionality, merge modules

[bugadani]

Multi-core operations, as well as CPU reset are no longer stable. CPU-related operations are now contained in the cpu module (as well as system reset, which may be slightly awkward but keeping that alone in a separate reset module seemed equally weird).

This PR also fixed CPU reset. In esp-idf, the rom function is defined as void esp_rom_software_reset_cpu(int cpu_no); and we were missing the cpu_no parameter. Because of this change, the function is no longer marked as stable.

I think these were the last bits in esp-hal that weren't supposed to be stable.

[MabezDev]

Sorry to bikeshed, I feel like this would be better placed in a system or soc module.

Which also brought my attention to https://docs.esp-rs.org/esp-hal/esp-hal/0.23.1/esp32s3/esp_hal/system/index.html which probably should be hidden completely I think?

[bugadani]

Good shout, although that system module is equally pointless IMHO Radio clocks belong to maybe clock, RadioPeripherals(?) and PeripheralIter belongs to peripherals.

[MabezDev]

Good shout, although that system module is equally pointless IMHO Radio clocks belong to maybe clock, RadioPeripherals(?) and PeripheralIter belongs to peripherals.

Yep, I agree with all those points!

[bugadani]

I moved everything into system, and removed the EnumIter thing that was a perma-stable-public implementation of the doc(hidden) Peripheral enum.

The Peripheral enum can and should probably be generated based on data in the soc::peripherals module, but for now it should be fine as it is.

[bjoernQ]

LGTM

[MabezDev]

LGTM, just one question that I either can't tell from the diff or has some other effect I'm not considering