Migrate from os.path to pathlib

[seisman]

Migrate from os.path to pathlib.Path following this table: https://docs.python.org/3/library/pathlib.html#correspondence-to-tools-in-the-os-module.

Please note that:

if os.path.exists(fname):
   os.unlink(fname)

is equivalent to Path(fname).unlink(missing_ok=True).

assert os.path.exists(fname)
os.unlink(fname)

is equivalent to Path(fname).unlink(missing_ok=False) or just Path(fname).unlink().

Closes #1834.

[seisman]

Will merge the PR in 24 hours if no objections.

[weiji14]

Do you want to enable ruff's PTH rules in this PR as mentioned at #1834 (comment), or in a separate PR?

[seisman]

Yes, but enabling PTH rules result in many violations against the PTH123 rule.

Better to do it in a separate PR and also need to discuss if we want to replace open with Path().open().

[weiji14]

I'm seeing that a lot of the assert os.path.exists() lines have been removed, but not replaced with Path.exists(), possibly because there is a Path.unlink() directly afterwards that would raise a FileNotFoundError if the file was missing (see https://docs.python.org/3/library/pathlib.html#pathlib.Path.unlink), is that correct?

I think there are some cases where we do need to explicitly do assert Path.exists(), especially if the test name is something like test_file_exists. Otherwise it's not apparent that .unlink() actually implicitly checks if the file exists already.

[weiji14]

Yes, but enabling PTH rules result in many violations against the PTH123 rule.

We could also apply PTH, but ignore PTH123 for now?

[seisman]

I think there are some cases where we do need to explicitly do assert Path.exists(), especially if the test name is something like test_file_exists. Otherwise it's not apparent that .unlink() actually implicitly checks if the file exists already.

Added back in ee0ba18.

Yes, but enabling PTH rules result in many violations against the PTH123 rule.

We could also apply PTH, but ignore PTH123 for now?

In some cases migrating from open to Path.open makes sense, so I prefer to do it in a separate PR.