Feature version variant hash

[ruck94301]

This PR to calculate a "hash" or signature of the ~50 py-files in the LabGym package.

The main frame displays f'LabGym v{LabGym.version}', like 'LabGym v2.9.7', just like before.

The version-string reported via central logger to central logging receiver
is like '2.9.7 (aec48e5691dd35bdd08ebc8265122156)'

This behavior is consistent with preference stated by Henry on 1/8 in comments of #318, "LabGym version if running a customized LabGym"

[ruck94301]

I'm a little puzzled right now. It's like this is now failing
# Confirm the assumption that under pytest, cwd is LabGym package's
# parent dir (the repo dir).
assert os.getcwd() == os.path.dirname(os.path.dirname(LabGym.file))

I'm not sure yet how this could be a new source of trouble...

[ruck94301]

I’m done refining this PR now.
The CI exposed three things:

1. An existing myargparse unit test needed to be updated for —version output.
2. hash.py was doing an os.chdir() instead of a temporary chdir, so an assumption in test_load.py was no longer true. This led me to improve hash.py by employing a temporary chdir.
3. The hashing of py-files used an open(file, ‘r’). But on Windows specifically, the default encoding was incompatible with some py-file content, so this needed to be changed to open(file, ‘r’, encoding=‘utf-8’)

These things are now corrected.