Corrected DTR ceiling

[emileten]

• [ ] closes #xxxx
• [x] tests added / passed
• [ ] docs reflect changes
• [x] entry in CHANGELOG.md

This corrects a mistake introduced by https://github.com/ClimateImpactLab/dodola/pull/163. The functionality to cap non-polar DTR values had an error in its implementation that lead all polar values to be modified. The tests couldn't catch this because they were also wrong, using == to compare xarray.Dataset objects.

This PR :

• Corrects the boolean condition expression in the cap function to preserve polar values
• Corrects the cap function test by making use of xarray.testing.assert_equals and adds an additional test case to verify that values that should not be capped (because those are not out of range) are indeed not capped.

[brews]

When this merges @emileten, do you want to take this for a test with dodola:dev on larger data or should we turn around a new patched release?

[emileten]

@brews thanks for the review. I'd like to test it first with dev.