cache CI environment

[zacharyburnett]

This PR does the following:

• expand test matrix (4 Python versions x 2 operating systems)
• add .flake8 for Flake8 configuration
• apply formatting to codebase to satisfy ruff checks
• runs additional MacOS jobs on a weekly scheduled workflow

[zacharyburnett]

Several Python versions on Linux are failing due to small differences in test vs reference data:

[zacharyburnett]

@stscieisenhamer @Witchblade101 are the differences in test success per-Python-version a concern?

[Witchblade101]

@zacharyburnett Yes. @mfixstsci plans on investigating this.

[Witchblade101]

I suspect this has been resolved in the recent PRs from @mfix, @pllim, and @sosey. We should review this to make sure before the next release.

[pllim]

ruff might be a step too extreme... but ok. 😸

[zacharyburnett]

ruff might be a step too extreme... but ok. 😸

Good point, it does ask for a lot of changes; I'll take it out

[mfixstsci]

@zacharyburnett should we remove python 3.7 and 3.8 from the testing matrix? Maybe add 3.12 if that's an option!

[zacharyburnett]

Alright, I took out 3.7 and 3.8, and added 3.12-dev to the weekly tests (which can be started manually from the web page). We could also add 3.12-dev to the on-push tests as well, but that might be more unstable

[mfixstsci]

Nice! Thank you @zacharyburnett. We can leave 3.12 out for now, I didn't know how far along it was in development. We can add it once there is a more stable version available. I don't know what it was about python 3.8 and causing our testing matrix to fail. It was the only version that has given us any trouble...