Development: ULP and ADC improvements

[purewack]

Hello, I was writing updates to improve the ADC driver so that the RISCV ULP can use the ADC driver in micropython as described in my PR's #16521 and #16630. As pointed out by @dpgeorge, this requires some back and forth with the community, as I have made some breaking changes to the ULP api.

I am happy to revert the functions that I have changed to comply with the existing ones, however I made these changes because I saw redundancy in the functions being called.
What I mean specifically is the loading and the running the loaded ULP code. I have merged the two into a single function as usually users would (most probably) load and then immediately run their ULP code, and if the code loading was to be delegated to a later time, simply loading at the required time shouldn't be too taxing as ULP binaries are very small, right?

I am posting this as I think the ULP (especially the RISCV variant) could be a useful feature for micropython users.

[jonnor]

Very cool to see these co-processors get some love.

Breaking the existing API in order to remove a single function call does not seem worth it. We have no idea how many programs are using the API as it is right now. And does not seem to be anything actually broken in the existing approach. If anything, the current way is the one that lets the Python side have more control, which is generally the preferred approach in MicroPython.

[Josverl]

I would suggest to keep the now redundant API for "a few" minor versions, but declare them deprecated. (Unless there is a significant size penalty to do so)

I've been adding @deprecated decorators in the type stubs to help flag the use of such APIs

[purewack]

There is now an effort to keep the compatibility as we don't know how many users rely on the ULP module as it stands.

There was an idea to create a separate class called RVULP to distinguish the different ULP types and maximize compatibility, as the current approach would allow to run compatible python code but error out at runtime if the user tried to run existing FSM binaries on RISCV

[Josverl]

The best way to prevent runtime errors is to warn developers as they are writing the code, preferably before they hit Run (aka inner loop)

For files with different types of information , the OS universal conventions is to use a different extension.
So foo.riscv is clear, foo.bin less so

with open("lowpower.riscv","rb") as file:
    riscv_prg= file.read()
ulp.run(riscv_prg)