Merge broadcast_like docstrings, analyze implementation problem

[DavidMertz]

This PR fixed the forked code connected to #3129.

• Closes New .broadcast_like() incorrect #3129
• Tests added
• Fully documented, including whats-new.rst for all changes and api.rst for new API

[pep8speaks]

Hello @DavidMertz! Thanks for updating this PR. We checked the lines you've touched for PEP 8 issues, and found:

There are currently no PEP 8 issues detected in this Pull Request. Cheers! 🍻

Comment last updated at 2019-07-15 14:37:22 UTC

[codecov]

Codecov Report

Merging #3130 into master will decrease coverage by 0.28%.
The diff coverage is 100%.

@@            Coverage Diff             @@
##           master    #3130      +/-   ##
==========================================
- Coverage   95.99%   95.71%   -0.29%     
==========================================
  Files          63       63              
  Lines       12799    12799              
==========================================
- Hits        12287    12250      -37     
- Misses        512      549      +37

[dcherian]

Thanks @DavidMertz

Line 2042 of dataset.py should be
return _broadcast_helper(args[1], exclude, dims_map, common_coords) i.e. args[1] instead of self.

Can you make that change (and the same change in dataarray.py, Line 1019)?

EDIT: I made the dataarray change directly.

[DavidMertz]

Made changes suggested by @dcherian. This resolves the detected bug in the test. If CI passes, this should be ready to merge. Removed the temporary method .broadcast_like_naive() and the test using it.

[dcherian]

Thanks for catching that @DavidMertz. Can you add back the original, simpler test too? This one is really just testing align. You could use @pytest.mark.parametrize to provide two different sets of arr1, arr2 arguments.

[dcherian]

LGTM.

[shoyer]

thanks @DavidMertz and @dcherian !