Expand testing of parse_access_ncfile to include correctness checks

[marc-white]

Is your feature request related to a problem? Please describe.

The current version of tests/builders/test_builders checks BaseBuilder.parse_access_ncfile against a fixed set of files and expected returns, i.e., regression tests it - however, it doesn't check the available file to see if the data returned are actually correct.

Describe the feature you'd like

Given we have access to the test files (and they're small, so no particular IO penalty), we could expand the test to check the correctness of the parse_access_ncfile return. The idea would be to look at each element of the parse_access_ncfile return, manually extract the related information from the test file, and compare them.

Ideally, this should be done in a slightly different way to how parse_access_ncfile does it, in order to provide a redundant sanity check.

Describe alternatives you've considered

Nil.

Additional context

I created an issue (#186 ) that suggests we should change the return of parse_access_ncfile from a tuple to a dict - doing so would make implementing this feature request easier.

[charles-turner-1]

@marc-white are you working on this actively/ have you made any progress on this? Thinking of picking this up - I'd like to expand the test suite before continuing with #153.

[marc-white]

@charles-turner-1 no I'm on other things at the moment, so feel free.

[charles-turner-1]

I've added a bunch of tests related to this on branch 187-correctness-checks.

As it stands:

• Updating intake-esm from 0.7 (current pinned version) => to 2.0.7 (most recent) seems to have no impact on results.
• Added tests are extensions of test_builders/test_parse_access_ncfile.
• Of the added tests, we get a failure on $ROOT/tests/data/access-om2/output000/ocean/ocean_grid.nc when we use a version of intake-esm that doesn't support searching for coordinate versions (eg. the most recent version, 2024.2.6). All other tests pass. The source of the error is expected & easily understood, but going to be a pain to reliably identify ahead of time so we can tell the test whether to xfail or not.

I don't think these changes are ready for a (draft) PR yet, so I'm leaving this open & pre-draft for now.

Next steps:
We need to extend our testing framework to test against

1. Different versions of intake-esm (source of the error)
2. Different versions of intake.

The obvious way to get this working is to use tox to test against the multiple dependency versions, so I'm going to get that configured and once we have that working as expected I'll open a PR.