Future neatening/readability

[truth-quark]

This issue is a general listing of tasks for cleaning up, refactoring & otherwise neatening the code base. The intent is to capture smaller tasks for future separate issues & PRs. One specific use case is deferring refinement & refactoring steps when PRs are growing too large. In those cases, feel free to defer non-essential fixes here while addressing core requirements.

Future tasks:

• Add module level docstrings (mostly for um2nc)
  • Can missing module docstrings be flagged as a warning in IDEs? (Pycharm & VS Code)
• Rename nonsnakecase variables to now_snake_case
• Convert old style % formatting to f-strings
• Fix/reformat long strings/PEP8
• Refactor repeated inline strings to named "constants"
• Refactor magic numbers to named "constants"
• Move set history to end of process() to group sman I/O operations (for future fixes)
• Move functions into call order (e.g. process() at top, then all the sub calls)
  • Arrange sub functions into sensible order
• Move set CF conventions to end of process()
• Move CLI code to another module?
  • Or better, move utility functions to other modules, shrink the core of um2netcdf.py
• Review/remove outdated TODO tags
• MuleVars renaming, see Refactor process() testing & cleaning cube masking/filtering logic #58 (comment)
• Reduce code duplication in fix_lat/lon_coord_name() functions
• Add lookup dict for Marc White's stash code suggestion
• Better testing assert error messages

These can likely be addressed during the unit test retrofit/refactoring step, on a function by function basis.