Python linting with Ruff

[jwallwork23]

Python linting with Ruff

Fixes #891

This PR adds ruff for Python linting, which is used widely by the community.

Useful info

• With ruff check we can see the current issues.
• With ruff check --fix we can get ruff to automatically fix many issues that are guaranteed not to break things.
• There is also a ruff check --fix --unsafe-fixes option but it should be used with care.
• There is also ruff format but it's quite opinionated so would likely completely reformat everything.
• ruff.toml contains the configuration, including any rules to ignore.
• The vast majority of the lines changed are in test/era5_topaz4_test_data.py because this is an enormous file.
• At some point in working on this PR, running ruff check --statistics gave the following:

58	D103   	undocumented-public-function
52	PLR2004	magic-value-comparison
27	D100   	undocumented-public-module
11	S101   	assert
 7	PLR0911	too-many-return-statements
 4	PLR0912	too-many-branches
 3	ARG001 	unused-function-argument
 3	PLR0915	too-many-statements
 2	E722   	bare-except
 2	PLR0913	too-many-arguments
 2	S602   	subprocess-popen-with-shell-equals-true
 1	A001   	builtin-variable-shadowing
 1	B007   	unused-loop-control-variable
 1	S607   	start-process-with-partial-path
Found 174 errors.

To get things to work, I applied ruff fixes, made code changes, and ignored rules that'd require more work to satisfy. The ignored rules are:

• Magic value comparison
• Too many statements/branches/arguments/return statements
• Subprocess call with shell=True
• Start process with partial path
• Missing docstrings for public modules/functions

I also applied per-file ignores to some of the code generation code for use of assertions as they'd need some refactoring to remove them and retain checks.

Checklist

• Make ruff check pass by fixing issues and ignoring rules, as appropriate.

[jwallwork23]

[Rebased on top of develop, excluding large data generation scripts from linting.]

[timspainNERSC]

Is there a reason the prescriptive second person form (generate), rather than the descriptive third person form (generates)? Admittedly I got this from writing javadoc comments many years ago, rather than any Python standard, but I feel it makes more sense to me.

[jwallwork23]

It comes from D401 First line of docstring should be in imperative mood. We could ignore this rule if preferred?

[timspainNERSC]

No, I'll try to follow that rule in future.

[timspainNERSC]

Big, but not complex. I can only apologize for the amount of trailing whitespace.