Use pooch instead of bundling standard name table

[dcherian]

Closes #422

cc @Zeitsperre

[codecov]

Codecov Report

Merging #436 (89f1321) into main (2d579f2) will increase coverage by 58.97%.
The diff coverage is 100.00%.

@@             Coverage Diff             @@
##             main     #436       +/-   ##
===========================================
+ Coverage   37.35%   96.33%   +58.97%     
===========================================
  Files          21       21               
  Lines        3916     2839     -1077     
  Branches      194        0      -194     
===========================================
+ Hits         1463     2735     +1272     
+ Misses       2259      104     -2155     
+ Partials      194        0      -194     

Flag	Coverage Δ
mypy	?
unittests	96.33% <100.00%> (?)

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
cf_xarray/accessor.py	94.81% <ø> (+58.84%)	⬆️
cf_xarray/tests/__init__.py	65.95% <100.00%> (+46.80%)	⬆️
cf_xarray/tests/test_accessor.py	99.80% <100.00%> (+70.62%)	⬆️
cf_xarray/utils.py	88.57% <100.00%> (+58.57%)	⬆️

... and 18 files with indirect coverage changes

[malmans2]

Looks good!
I would just change the docstring of argument source in add_canonical_attributes, to something like:

If None, use the latest CF standard name table.

[malmans2]

Another quick comment, if we use fsspec rather than pooch, I think it's more likely that it is already installed as it should be a dependency of dask.
See: #422 (comment)

[Zeitsperre]

@dcherian Thanks for the heads up!

[dcherian]

if we use fsspec rather than pooch, I think it's more likely that it is already installed as it should be a dependency of dask.

I think its OK. Xarray uses pooch for its tutorial datasets anyway...