STM32: Configurable bank support

[diondokter]

It builds locally, but not sure how I can make the metapac work here without it being merged there since the stm32-data-generated only exists for main...

embassy-rs/stm32-data#596

To make this work, I added two features 'single-bank' and 'dual-bank'. They are optional. If a chip can't configure this, the user doesn't have to select one.
If a chip does have multiple flash options, selecting one of the features is required. There's also a cfg being set so the flash drivers knows it needs to check the configured settings.

The config system is generic, but it's only implemented for the G4 in this PR.

An alternative approach is going the same route as the dual core settings where you get a STM32* feature for single bank and one for dual bank. Let me know if that's more desirable.

closes #2309

[diondokter]

Alright, I've implemented the configurable flash support for all STM chips I could find that have that feature in one way or another

[diondokter]

bender run

[embassy-ci]

run: permission denied

[jamesmunns]

bender run

[bugadani]

bender run

[embassy-ci]

run: permission denied

[jamesmunns]

Fake approval to see if that makes bender happy, don't actually merge

[jamesmunns]

bender run

was just a test

[diondokter]

Alright, this should be finished.

I found there was some dual bank implementation for the F4 series. So I could rip that out. This makes the PR net negative LOC! :D

The F4 & F7 series are weird with their banks and sectors, but they should work.
For anything that runs on the HIL I've added dual/single-bank feature with which they should be factory programmed, so should all remain working.

[jamesmunns]

bender run

[diondokter]

Hmmm well, I'm sure I didn't break the uart and that it's flaky somehow

[lulf]

Looks good to me, I've merged embassy-rs/stm32-data#596 so please update this PR with the correct ref.

[diondokter]

Alright, that should be it!