Add KGE Score

[tennlee]

This incorporates the work done by @durgals to add the KGE score (Issue #678, PR #679) and additionally includes some changes I applied in response to the code review on PR #679.

@durgals the test coverage is not 100%. I have included a screenshot showing which lines do not have test coverage. I have also pushed this work to the 228 documentation branch (which we use to test the documentation) Please go through the checklist below and check each thing off.

• Lines (523-528) are the only ones which do not have test coverage. Please add unit tests covering these conditions and assert that the appropriate exceptions are raised.
• Please visit https://scores.readthedocs.io/en/228-documentation-testing-branch/included.html and check this looks right for KGE
• Please visit https://scores.readthedocs.io/en/228-documentation-testing-branch/api.html#scores.continuous.kge and check this looks right for KGE
• Please visit https://scores.readthedocs.io/en/228-documentation-testing-branch/tutorials/Kling_Gupta_Efficiency.html and check this looks right to you also

Also, if you are unhappy with any of the changes I made, please feel free to say how it should be changed.

[tennlee]

@durgals @nicholasloveday

Thanks for that PR Durga. I have merged it into this PR. So far as I know, everything is now ready to be merged. I have pushed the updates to https://scores.readthedocs.io/en/228-documentation-testing-branch/ .

Could you both please take a final look and let me know with a short message here? If you're both happy I will go ahead and merge this into develop.

Note - rather than squashing this into one single commit, I will probably organise it into two commits, but I will take care of that as part of the merge process - neither of you need to do anything about it.

[nicholasloveday]

I think that this is fine to merge in. Thank you for this work.

A couple of minor things to mention.

• Consider removing the dots in the KGE equation (in case someone thinks that this is the dot product). I am still okay if this is merged in as is.
• In the future, I think that we should try to provide more support for supporting xr.Datasets as inputs. No changes needed to this at this stage.
• Did you want "are the scaling factors for the correlation coefficient" bold in the docstring?

[tennlee]

I have merged this into the develop branch now. @durgals can you please review @nicholasloveday 's final comments and consider if you would like to raise new issues to track them. I'm fine either way in each case.

[durgals]

@tennlee Thank you merging this. Regarding comments from @nicholasloveday:

• Consider removing the dots in the KGE equation (in case someone thinks that this is the dot product). I am still okay if this is merged in as is.
  I will update this in the next round of changes, including support for xr.Dataset or weights.
• In the future, I think that we should try to https://github.com/nci/scores/issues/702for supporting xr.Datasets as inputs. No changes needed to this at this stage.
  I’ll consider this for a future release,
• Did you want "are the scaling factors for the correlation coefficient" bold in the docstring?
  I already reported this to @tennlee, and he mentioned that it’s a rendering issue with ReadTheDocs.