Refactor arco

[jennan]

This PR implements a refactoring of the arco module, with the following changes:

• remove dead code (commented out or unused helper functions) and simplification remove the need for helper function (i.e. pre-generate the inverse mapping from short variable names to long variable names)
• move all code and validation variables in one module, as the simplifications make the code more compact and still readable in one module with less than 500 lines
• accelerate loading of few variables by dropping all other variables in open_zarr call
  • loading the full dataset takes (all levels) takes 1min30s
  • loading 1 variables (all levels) takes 2s
• drop of the global variable singleton approach
  • tend to avoid global variables as they create side-channels not visible for a user and I fell this is not needed if the goal was to make loading faster, in particular in a case of loading few variables
  • the original implementation wasn't re-opening the dataset if the kwargs changed
  • this could be replace by caching (either functools.cache or joblib.Memory) if really needed, knowing that the full dataset takes about 18GB of memory once opened

Missing elements to complete to get the PR ready:

• add tests
• add documentation at the module level

[coveralls]

Pull Request Test Coverage Report for Build 15316217872

Details

• 93 of 98 (94.9%) changed or added relevant lines in 4 files are covered.
• No unchanged relevant lines lost coverage.
• Overall coverage increased (+0.3%) to 61.117%

---

Changes Missing Coverage	Covered Lines	Changed/Added Lines	%
packages/data/tests/download/test_arcoera5.py	51	53	96.23%
packages/data/src/pyearthtools/data/download/arcoera5.py	39	42	92.86%

Totals
Change from base Build 15313836174:	0.3%
Covered Lines:	9937
Relevant Lines:	15805

---

💛 - Coveralls

[jennan]

@tennlee I have added tests and the docs to the API docs. The PR is ready :-).

[tennlee]

I have taken a brief look, the direction seems fine. I will spend some more time with it over the next few days and may have some additional feedback based on that.

[jennan]

@tennlee I think I have addressed the only action point following our discussion, i.e. renaming the pytest mark. Regarding the failing pre-commit, I don't manage to trigger it locally and it is related to a file I have not touched, packages/tutorial/src/pyearthtools/tutorial/ERA5DataClass.py so let me know if this should still be fixed as part of this PR.

[tennlee]

Looks good. I'll investigate the pre-commit hook separately.