Fix Windows build and lint

[zanieb]

@benfred regarding #679 (comment)

Includes a few other changes as we debugged the cause of the failure.

[benfred]

Thanks for the PR! The changes look good to me,

[benfred]

I'd like to get the windows CI working before merging - I realized I might have messed up the how python is installed in the last PR (moved setup-python to the wrong location in the GHA yaml), and am also testing out a tweak to fix in this other PR
#642

[zanieb]

Yeah agree. I'm pulling updates from my fork to see if one of those fixes the issue — I don't know what else I would have changed to make CI pass.

[zanieb]

It looks like you might have changed from Python 3.9 -> 3.11?

[benfred]

I'm sorta wondering if this is related to the python version being tested - I'm getting a different error with python 3.10 than with python 3.11 in CI, and I'm wondering if the working versions are with python 3.9.

Because of how I messed up the setup python install, its possible that the 'fix clippy' pr we merged tested with python 3.9 (since the install for python 3.11 happened afterwards).

[benfred]

It looks like you might have changed from Python 3.9 -> 3.11?

yeah, I'm testing now on #642 with 3.9 to see.

The error with 3.10 was called Result::unwrap()on anErrvalue: Failed to find a python interpreter in the .data section https://github.com/benfred/py-spy/actions/runs/11279360838/job/31369936001?pr=642 - which might be the same as #684 (so potentially a real issue with python 3.10 on windows, rather than a false positive in CI)

[zanieb]

Yeah 1d0ffc1 resolves the error.

If you want, I can separate that commit from this pull request. Otherwise, it seems fine to merge this as a whole.

[benfred]

I'm good merging this as a whole - thanks for fixing.

It does like there is an issue on windows with python 3.10+ that needs fixing =( - but that doesn't need to be fixed in this PR.