Heavy spi module refactor

[mjsir911]

Takes advantage of existing kernel procedures for gpio handling mostly, some resource management and device loading too.

Oh, this actually doesn't work because:

• devm_spi_register_master does automatically allocation CS pins, but looks them up in a table with a device it literally just named (??)
• I made a patch to move the naming to happen after the table gets created and have been working with that patch locally.
• Also sent an email out to the mailing list for help but got back radio silence.

Also regresses the spi driver behaviour by claiming all CS pins / spi busses immediately. It would be cool to be able to load/unload individual busses individually (like we've previously been able to do), if there's some convention on how to do that within the kernel that would be a good start. This probably also means not relying on the kernel's csgpiod handling as much.

[mjsir911]

Okay yeah I got clued in about multiple devices and spi does fail to load on the second device plugged in.