Fix for xESMF 0.6.1

[aulemahal]

Pull Request Checklist:

• [x] This PR addresses an already opened issue (for bug fixes / features) - no related issues

• [x] Tests for the changes have been added (for bug fixes / features)

• [x] Documentation has been added / updated (for bug fixes / features) - this is not needed as the change is minor and is retrocompatible.

• [x] HISTORY.rst has been updated (with summary of main changes)

• [x] I have added my relevant user information to AUTHORS.md

• What kind of change does this PR introduce?: Minimal change to adapt to new xESMF 0.6.1. In that version, the weights attribute of the Regridder object is a DataArray, instead of a scipy sparse matrix. The change is retrocompatible.

• Does this PR introduce a breaking change?: No

• Other information:

[huard]

Are the non-implemented errors expected ?

[aulemahal]

Not at all and I can't see how my PR would have triggered those?

[huard]

@ellesmith88 Any idea on what could be causing this ?

[ellesmith88]

@aulemahal @huard From a quick look, this is to do with the newer releases of pandas and xarray. This error appears when using pandas>=1.3.0 and xarray>0.18.2. Seems to be to do with this PR https://github.com/pydata/xarray/pull/5359

Also see https://github.com/roocs/clisops/issues/185

[ellesmith88]

@aulemahal @huard The suggested workaround for this for now is https://github.com/roocs/clisops/commit/c21275fd962ec7fbf48ec28c70b422a5bb37da11.

I've tested the fix locally in your branch and all tests passed for me. The fix is in a clisops branch - dask-error-workaround ( https://github.com/roocs/clisops/tree/dask-error-workaround), if you merge this in the issue should be resolved.

The PR looks good otherwise.

[huard]

Thanks !

[agstephens]

Should be ready to merge. @agstephens double check.

[agstephens]

@aulemahal please add a line into HISTORY.rst to record the change - then we can merge this to master. Thanks

[agstephens]

Thanks @aulemahal