Loosen check on coordinate class for NDData translator to accept subclasses

[bmorris3]

The NDData translator checks that the Data.coords is either a WCS object, a glue Coordinates object, or None. Other Coordinates types are not allowed in the current logic. I'm trying to work with a case where the coordinate attribute that's an IdentityCoordinate, but that's disallowed.

This PR simply allows coords attributes that are subclasses of Coordinates.

[dhomeier]

Thanks, this looks reasonable enough, and since you introduced the translator, I trust there are no issues to be worried about! Would it be possible, still, to add this as test case e.g. to test_to_ccddata?

Also, depending on whether you consider the previous refusal of subclasses a bug, this should be labelled as bug or enhancement for the automatic changelog.

[bmorris3]

@dhomeier I've added a test. I'd consider this a bug. It makes sense for CCDData to enforce the coordinates to be relevant to observations from a CCD. NDData shouldn't make the same assumptions about the class used to define coordinates.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 96.71%. Comparing base (5c91529) to head (a2bfe9a).
Report is 1 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #101      +/-   ##
==========================================
+ Coverage   96.69%   96.71%   +0.02%     
==========================================
  Files          18       18              
  Lines        1453     1462       +9     
==========================================
+ Hits         1405     1414       +9     
  Misses         48       48              

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[dhomeier]

Thanks! Could you rebase to get the other tests running?

[dhomeier]

Everything green, thanks!