Force gwcs to always return a F ordered bounding box

[WilliamJamieson]

This PR forces GWCS to always return an F ordered bounding box. It raises a warning when it discovers a non-F ordered bounding box that it needs to convert.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 87.42%. Comparing base (eb9d316) to head (55223f9).
Report is 23 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master     #522      +/-   ##
==========================================
+ Coverage   87.28%   87.42%   +0.14%     
==========================================
  Files          22       22              
  Lines        3821     3874      +53     
==========================================
+ Hits         3335     3387      +52     
- Misses        486      487       +1     

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

[nden]

jwst regression tests: https://github.com/spacetelescope/RegressionTests/actions/runs/11733040618
romancal regression test: https://github.com/spacetelescope/RegressionTests/actions/runs/11733050524

[pllim]

This warning is very disruptive. It is getting emitted in Jdaviz for just existing. For example:

jdaviz/configs/imviz/plugins/parsers.py:238: in _parse_image
    if isinstance(data.coords, GWCS) and (data.coords.bounding_box is not None):

Here, we merely read in the GWCS from an existing JWST ASDF-in-FITS image and checking whether it is set. Not even doing anything with it. Bam, warning. What are we supposed to do here? We cannot really go change the file in MAST. Hope you can advise.

[pllim]

You mean specifically astropy.modeling, right? Things might be different in astropy.wcs. Should clarify.

[WilliamJamieson]

astropy.wcs has no notion of a bounding box.

[pllim]

I am still confused. If downstream, we use it like this, are we already using it correctly? https://github.com/spacetelescope/jdaviz/blob/5130cafd8f8790a3e09b2d82c5ee721a1b909d9f/jdaviz/configs/imviz/wcs_utils.py#L282-L292

        ints = data.coords._orig_bounding_box.intervals
        if isinstance(ints[0].lower, u.Quantity):
            bb_xmin = ints[0].lower.value
            bb_xmax = ints[0].upper.value
            bb_ymin = ints[1].lower.value
            bb_ymax = ints[1].upper.value
        else:  # pragma: no cover
            bb_xmin = ints[0].lower
            bb_xmax = ints[0].upper
            bb_ymin = ints[1].lower
            bb_ymax = ints[1].upper

[pllim]

If I have to extrapolate from the warnings I already see over at specutils and jdaviz , the way the warning is emitting now, you might get a lot of questions from users just trying to read existing JWST files out there when this patch is released. FYI.

[pllim]

Just so I can find the trail in the future:

• Ignore the bounding box ordering warning when reading files #526