Update docs to reflect open_mfdataset default chunk behaviour

[jemmajeffree]

Looking at #5704 and #9038 (and various other issues that reference these), it seems like there's no easy solution and the current open_mfdataset chunk default is going to stay for a while. i.e. chunks=None as a kwarg to open_mfdataset is overwritten with chunks={} before calling open_dataset, so the default chunks are disk chunks not single-file chunks.

xarray/xarray/backends/api.py

Line 1677 in 5ce69b2

open_kwargs = dict(engine=engine, chunks=chunks or {}, **kwargs)

In which case, we should probably update the docs to reflect this current behaviour.

[welcome]

Thank you for opening this pull request! It may take us a few days to respond here, so thank you for being patient.
If you have questions, some answers may be found in our contributing guidelines.

[jemmajeffree]

@dcherian would you mind taking a quick look at this for me?
It's unclear to me while some of the tests are failing, given the only thing I changed was a docstring

[keewis]

I would expect the numba CI failure to either be unrelated or just flaky (and we don't have to care about the PR labeler here). I've restarted the CI, let's see if that changes things.

Edit: all CI passes!

[jemmajeffree]

Thanks! Anything else I should do here, or am I just waiting for someone to check the docstring update is correct and merge?