Introduce multi-variable map plotting

[ukmo-huw-lewis]

Addresses #1580.

Contribution checklist

Aim to have all relevant checks ticked off before merging. See the developer's guide for more detail.

• Documentation has been updated to reflect change.
• New code has tests, and affected old tests have been updated.
• All tests and CI checks pass.
• Ensured the pull request title is descriptive.
• Conda lock files have been updated if dependencies have changed.
• Attributed any Generative AI, such as GitHub Copilot, used in this PR.
• Marked the PR as ready to review.

[github-actions]

[ukmo-huw-lewis]

Updates include support for calling from workflow, and command-line testing after merge with latest main.

To do:

• add tests
• resolve issues / inefficiency with ambition to collapse required cubes (may need to be deferred to future issue/PR).

Example output (filename lfric_multi-variable_grid_surface_temperature_seq_20241005230000.png):

[ukmo-huw-lewis]

Changeset updated to
a) remove time-averaging elements from this PR (note the workflow 'plumbing' are still in this PR, but will have no impact).
b) add test of new routine.

On the time-averaging aspects, will setup new issue/PR to cover these elements, as think might have benefits/implications across other supported recipes for general capability rather than only specific to this recipe.

Ready for review.

[James Warner (jwarner8)]

Testing over SAfrica shows this overlay working nicely. Will finish review shortly

[James Warner (jwarner8)]

Looks good, few minor suggestions. Highly useful functionality that will need broadening in future (e.g. transects, or aggregation), but advantageous to get this on trunk now.

[ukmo-huw-lewis]

Thanks for review comments. See responses above.
Main changes now implemented:

a) Adopted proposed approach for slice_over_maybe, and seems to preserve required behaviour.
b) Implemented suggested change to add contour labelling. Worthwhile to add quantitative information.
c) Implemented suggested change to add cbar for the overlay variable. Avoided doing this initially to preserve focus on map plot rather than proliferate cbars, but think worthwhile for clarity of what is being plotted (and more publication-ready default). Have checked that no impact on regular (1-variable) spatial plotting.

Example updated output following revisions:

[James Warner (jwarner8)]

Latest branch run;

[ukmo-huw-lewis]

Ahead of merge to main, I ran a ensemble workflow as struggled testing with ensembles on command-line cset bake.
Unfortunately tasks failed on reading, linked to #1972. Will update the relevant issue.

As per discussion above, proposal is to merge this PR to main, with unit testing covering ensemble use-case. Previous testing in development of this change illustrated ensemble multi-variable outputs working correctly. Can review again following improvement to ensemble read reliability.

Will merge this PR as-is.