Prevent astropy warnings in tests

[eslavich]

astropy 4.2.1 started emitting an additional warning when opening bad files. This PR marks tests that open known bad files to ignore that warning.

Resolves #966

[jdavies-st]

Do we know why these are bad files? Almost every test file that the asdf library has written now gets these warnings. There are dozens in the jwst repo as well. 🤔

[eslavich]

Do we know why these are bad files?

In asdf at least they're intentionally bad, for example:

def test_no_asdf_header(tmpdir):
    content = b"What? This ain't no ASDF file"

That content isn't recognized as ASDF, so it gets passed to astropy.io.fits, which emits the warning. Do you have examples in jwst of files that should be valid ASDF or FITS but still result in a warning?

[jdavies-st]

Yeah, a ton of them. Start here and scroll up.

https://github.com/spacetelescope/jwst/runs/2422915561?check_suite_focus=true#step:10:845

That said, you bring up a good point. Some are likely bad files, but I've looked at a few, and they are not even writing out files, which makes me think AsdfInFits is generating bad files? Not sure. Would be good to understand whether it's a bug in astropy.io.fits or a bug in asdf.AsdfInFits instead of sweeping it under the rug.

[eslavich]

This commit seems responsible:

astropy/astropy@070cd94#diff-6ebde8ab98c018da8dd092e05ef30d9607dc9f998bfed0df91a6b67c671ca651

The warning is emitted from within that _check_padding function. Previously they were raising "Header missing END card." for invalid files before _check_padding. Now the order is reversed so the warnings are appearing before the exception is raised.

[jdavies-st]

This looks good.

[jdavies-st]

And let's discuss any issues with jwst elsewhere, preferably in a jwst issue.