Include metadata variables for (arrays of) structures in to_xarray

[maarten-ic]

This allows reading back data in the following scenario:

ids = imas.IDSFactory().core_profiles()
... # fill IDS
xrds = imas.util.to_xarray(ids)
xrds.to_netcdf("data.nc", group="core_profiles/0")

with netCDF4.Dataset("data.nc", "a") as ds: 
    # Set global DD version metadata:
    ds.data_dictionary_version = imas.util.get_data_dictionary_version(ids)

with imas.DBEntry("data.nc", "r") as entry:
    ids2 = entry.get("core_profiles")

[maarten-ic]

N.B. I want to further close the gap between writing an IDS to a netCDF file (with DBEntry.put) and the xarray route. Marking this PR as draft until then.

[maarten-ic]

@prasad-sawantdesai @olivhoenen I've updated the logic so that:

• An IDS written to netCDF and loaded with xarray.load_dataset equals to an ids converted to xarray with imas.util.to_xarray()
• An IDS converted to xarray and then store to netCDF can be read with DBEntry.get() (see example in description above)

Can you please review the PR?

Thanks!

[olivhoenen]

Works nicely, thanks @maarten-ic !
Could we describe the step

with netCDF4.Dataset("data.nc", "a") as ds: 
    # Set global DD version metadata:
    ds.data_dictionary_version = imas.util.get_data_dictionary_version(ids)

somewhere in the documentation (maybe after the section on converting directly IDS to xarray)?

[maarten-ic]

Works nicely, thanks @maarten-ic ! Could we describe the step

with netCDF4.Dataset("data.nc", "a") as ds: 
    # Set global DD version metadata:
    ds.data_dictionary_version = imas.util.get_data_dictionary_version(ids)

somewhere in the documentation (maybe after the section on converting directly IDS to xarray)?

Thanks for the suggestion! Please check if the documentation is clear like this: https://imas-python--112.org.readthedocs.build/en/112/netcdf.html#store-xarray-datasets-in-imas-compatible-netcdf-file