Fix macOS wheels: drop setuptools-scm

[crwood]

This PR removes the use of setuptools-scm in favor of declaring a hardcoded version string, effectively fixing the failing (macOS) wheel builds. Builds had been failing previously because newer versions of setuptools-scm can no longer be configured via setup.py (and must use pyproject.toml).

This is not ideal solution: in keeping up with recent changes in the Python packaging ecosystem, it would be preferable to shift project configuration from setup.py to pyproject.toml, however, this project's setup.py currently contains custom/milksnake-specific hooks that would presumably be need to re-implemented elsewhere (and is beyond the scope of this PR).

Note that a version string was already being hardcoded in nix/python-challenge-bypass-ristretto.nix (so setuptools-scm could not have been used to handle any/all version-string declaration anyway). I've documented this in a comment in both files; future maintainers should update the version string in both places until the entire project can be migrated to using pyproject.toml.

[hacklschorsch]

Looks good to me!

Where can I see the difference in macOS wheels not working before and now working?
I thought I had already fixed this, else I hadn't bumped the version in my earlier PR :/

[crwood]

Where can I see the difference in macOS wheels not working before and now working?

I had tried building locally and got the failure shown in #58 (which you could also see here -- albeit under ZKAPAuthorizer).

I thought I had already fixed this, else I hadn't bumped the version in my earlier PR :/

Oh, that's interesting. I hadn't noticed that your change passed the macOS builds independently at the time I started this (probably because it was red at the time that we merged -- but for different reasons). I'm still not sure I understand how/why setuptools-scm "Just Works" there and not here, but I'm glad that it does.

Feel free to reject/close this (and, I guess, close #58?). Sorry!

[hacklschorsch]

Should we still merge this now that the wheels build fine without it?

[crwood]

Nah, let's just close it. I still don't know what led to the original breakage here but we don't gain much by trying to diagnose it (plus, a lot of adjacent things were broken at the time so it could have been related to something that is no longer a cause).