Ensure test/example datasets are all CF compliant

[sadielbartholomew]

We've realised that some of the dataset files we use for the documentation examples and integration testing are not actually CF compliant - so they are really not good to showcase and test on.

So we should (post merge of the #123 big revamp) review all of the datasets we use for these and ensure they are not just fully CF compliant but good exemplars for the specific visualisation case (as summarised by the titles in the form 'Example N: <title>' listing in the 'Plots by type/theme' page of the docs).

It also wouldn't hurt to increase the diversity of datasets we showcase visualisation of with cf-plot (but that could be a separate Issue).

[sadielbartholomew]

David pointed out that a good and compliant test file for UGRID is available already as part of the cf-python test datasets: https://github.com/NCAS-CMS/cf-python/blob/main/cf/test/ugrid_global_2.nc. And I think we can draw on a lot more from cf-python (and/or cfdm) as a good way forward.

[sadielbartholomew]

I'd also like to contribute in the process to this (volunteered already but well overdue to actually start adding datasets): https://github.com/cf-convention/cf-example-datasets so let's aim to do that all at once.