refactor(sercom): clean up underlying pad data structures

[FraFrieFa]

This change lays the groundwork for implementing synchronous and half-duplex UART by cleaning the pad modules. Each commit compiles on its own.

• Introduce hal/src/sercom/pads.rs
  • A four-slot, type-level Pads<P0, P1, P2, P3> container, representing GPIO pins for one Sercom (each Sercom has up to 4 Pads: Pad0, Pad1, Pad2, Pad3 which can only be assigned to one Pad each -> suitable struct for storing underlying pins)
    • Holds up to 4 pads, Sercom is automatically determined
  • A macro roles_type to combine Pads with protocol roles (Rx, Tx, ...)
  • ValidPads trait
    • checks all pads share one Sercom (ShareSercom)
    • thumbv7em: also checks a common IO set (ShareIoSet)
• Replaced the implementation for Sercom protocols with the new Pads + Roles structures
• Pad access traits PadSet and ValidPads have been reduced to ValidPads
• Delete all thumbv6m/thumbv7em “pad” files for I²C, SPI, and UART
• Update all board BSPs and examples
  • Remove SercomX arguments in pad aliases (e.g. uart::Pads<Sercom0, …>)
  • All examples now have to convert all raw pins via .into() to pick the correct alternate function

TODO:

• Verify python3 build-all.py runs for each commit
• Verify that the new Pads behave like the old ones (no invalid configurations)
• Fix Clippy lints and ensure CI passes on all chips
• Consider ergonomic constructors to avoid manual .into() calls (the old code automatically converted a Pin into the correct Pad, but replicating that behavior requires specifying a Sercom) A DormantPads struct might hold GPIO pins until it is activated by claiming a Sercom slot (for example in the Uart constructor), coverting them to their alternate functions

This intermediate refactoring state is shared for review and to help the team become familiar with the changes.

[ianrrees]

Not directly about this PR (which, sorry, I've only skimmed over but so far it looks good!), but I've been thinking we should use tests to cover typelevel code, maybe using a technique like one of the answers to https://stackoverflow.com/questions/74989492/how-to-assert-that-code-wont-compile-in-a-test