Allow apply_ufunc to ignore missing core dims

[max-sixty]

• Closes apply_ufunc and Datasets with variables without the core dimension #6391
• Tests added
• User visible changes (including notable bug fixes) are documented in whats-new.rst
• New functions/methods are listed in api.rst

This probably needs a review and another turn, maybe some more tests on multiple objects etc. There are a couple of questions inline.

[max-sixty]

Any thoughts on this, team?

[mathause]

We may have to think about how we handle multiple input_vars and what happens if multiple ds are passed (e.g apply_ufunc(func, ds1, ds2)). But that are probably rather edge-cases.

[max-sixty]

What's the best name for this arg?

• missing_core_dim? Possibly too synonymous with other args which actually list dims, like exclude_dims?
• on_missing_core_dim? Arguably more fluent with on_missing_core_dim="raise", but less standard...

[max-sixty]

OK, I'm reasonably happy with this now — it's ready to go, pending docs & changelog. Any thoughts prior?

[max-sixty]

We may have to think about how we handle multiple input_vars and what happens if multiple ds are passed (e.g apply_ufunc(func, ds1, ds2)). But that are probably rather edge-cases.

Sorry, just saw this. Yes, I actually had a TODO prior re adding tests for this, which I just added.

In the case of drop & raise, the order doesn't matter. In the case of copy, we copy from the first argument. Do you think that's reasonable?