Rename chunks kwarg to partition_chunks into open_datatree method

[sjperkins]

The chunks keyword argument in open_dataset and open_datatree is reserved. In the case of open_datatree it is simply passed through to the open_dataset method.

Previously, xarray-ms allowed a custom per-partition chunking schema to be passed via the chunks kwarg in MSv2PartitionEntryPoint.open_datatree. For e.g.

dt = open_datatree("measurementset.ms", chunks={
  ("DATA_DESC_ID", 0), ("FIELD_ID", 0)): {"time": 2, "baseline": 2},
  ("DATA_DESC_ID", 0), ("FIELD_ID", 1)): {"time": 3, "baseline": 4}
}

While MSv2PartitionEntryPoint.open_datatree understands this chunking schema, other storage backends do not so there is potential for undefined behaviour.

This PR moves this functionality from chunks kwarg to the new partition_chunks kwarg which is only understood by MSv2PartitionEntryPoint.open_datatree. This separates xarray-ms's custom chunking behaviour from xarray's default chunking behaviour.

• Test Cases covering your PR.
• Documentation.
• A Changelog entry in doc/source/changelog.rst.

---

📚 Documentation preview 📚: https://xarray-ms--37.org.readthedocs.build/en/37/

[sjperkins]

@JSKenyon @landmanbester @o-smirnov Would you mind reviewing this general logic behind this change? The reasoning is explained further in

• Rename chunks kwarg to partition_chunks into open_datatree method #37 (comment)

[JSKenyon]

This doesn't look controversial to me, and it is better to change these things sooner rather than later.

[landmanbester]

Agree, seems uncontroversial and good to do this now rather than later

[sjperkins]

Thanks for the review.