Pint support for top-level functions (#3611)

* get the align tests to pass

* add pint to the upstream-dev ci job

* special case for booleans

* silence the pint behaviour change warning

* preprocess the unit mapping parameter to convert_units

* use assert_allclose and assert_identical instead

* clean up a few tests

* remove some xfails

* use the unit registry's quantity class

* explain the catch_warnings block

* don't use the function wrapper class if we don't need arguments

* whats-new.rst

* require the new pint version

* use functools.partial instead of function

* remove the convert_from parameter of array_attach_units

* make sure every top-level function test uses assert_units_equal

* hide the traceback of the unit comparison function

* considerably simplify the merge_dataarray test

* simplify the merge_dataset test

Author: keewis

ci/requirements/py36-min-nep18.yml

@@ -11,7 +11,7 @@ dependencies:
   - msgpack-python=0.6  # remove once distributed is bumped. distributed GH3491
   - numpy=1.17
   - pandas=0.25
-  - pint=0.9  # Actually not enough as it doesn't implement __array_function__yet!
+  - pint=0.11
   - pip
   - pytest
   - pytest-cov

doc/whats-new.rst

@@ -32,7 +32,7 @@ New Features
 - Support new h5netcdf backend keyword `phony_dims` (available from h5netcdf
   v0.8.0 for :py:class:`~xarray.backends.H5NetCDFStore`.
   By `Kai Mühlbauer <https://github.com/kmuehlbauer>`_.
-- implement pint support. (:issue:`3594`, :pull:`3706`)
+- Support unit aware arrays with pint. (:issue:`3594`, :pull:`3706`, :pull:`3611`)
   By `Justus Magin <https://github.com/keewis>`_.
 - :py:meth:`Dataset.groupby` and :py:meth:`DataArray.groupby` now raise a 
   `TypeError` on multiple string arguments. Receiving multiple string arguments