Improve testing

[valeriupredoi]

Hey @RosalynHatcher hope you don't mind me nosing a bit here 😁 I wrote a quick Github Actions test flow for you (with some explanations in comments) so you can test the installation from PyPi; you can write a similar one to test the install from source, for developers. A few notes from me:

• tfunctional testing can be improved, be careful that the current test script fails if you don't create the cache dir in a user-accessible location, also it'd be good to put those target dirs outside the git-controlled repo, so you don't create orphan files
• the nc files that the tests use get modified in-place so that's not good since they're git-controlled

Another side-note, cfchecks --version returns the help docstring but not the version, I think you are not setting it correctly at argparse stage, sorry, am nagging too much 😁

There is also an issue I wanted to flag: it's best you recommend users+developers (the advanced users) to install it in a virt env (either Python or Conda virt env), a la:

conda create -n cfchecker
conda install pip
conda install -c conda-forge udunits2  # this is a runaway dependency that doesn't get installed with pip install -e .
pip install -e .

This is better and more secure than python setup.py install

Anyways, hope this is useful 🍺

[valeriupredoi]

BTW here's the issue with udunits2 as picked up by the GA test without pre-install of it, installing it fixes the actual functionality but it looks like that your help returns a non-zero exit status, check it out, probably related to the issue with the version not being stdouted