ENH/BUG: avoid test pollution from data tests by making APIs resilient against missing cache directories - #19650

Thank you for your contribution to Astropy! 🌌 This checklist is meant to remind the package maintainers who will review this pull request of some common things to look for.

• Do the proposed changes actually accomplish desired goals?
• Do the proposed changes follow the Astropy coding guidelines?
• Are tests added/updated as required? If so, do they follow the Astropy testing guidelines?
• Are docs added/updated as required? If so, do they follow the Astropy documentation guidelines?
• Is rebase and/or squash necessary? If so, please provide the author with appropriate instructions. Also see instructions for rebase and squash.
• Did the CI pass? If no, are the failures related? If you need to run daily and weekly cron jobs as part of the PR, please apply the "Extra CI" label. Codestyle issues can be fixed by the bot.
• Is a change log needed? If yes, did the change log check pass? If no, add the "no-changelog-entry-needed" label. If this is a manual backport, use the "skip-changelog-checks" label unless special changelog handling is necessary.
• Is this a big PR that makes a "What's new?" entry worthwhile and if so, is (1) a "what's new" entry included in this PR and (2) the "whatsnew-needed" label applied?
• At the time of adding the milestone, if the milestone set requires a backport to release branch(es), apply the appropriate "backport-X.Y.x" label(s) before merge.

I don't think this is critical for v8.0.0 release, so moving milestone

Please don't. There's a subtle bug I'm still tracking down and plan to fix here, which I really don't want to hit users. It's not that critical for a beta but I plan to get it fixed before rc.

Thank you. I addressed points 1 to 4.
Points 5 to 7 also seem worth pursuing but require more time than I have right now.

Point 5 is actually nonesense

Changed if not (path).is_dir() → if (path).is_file(). This now lets a non-existent path through

This is simply false because is_file() implies exists()

6. Still holds as a valid critic. I would actually agree that some functions do not need to warn by default, but what I have now allowed me to discover more cases that previously relied on the assumption that querying a cache dir guaranteed its existence. Maybe we can change it here or in a follow up PR (possibly defered to RC ?)
   I think a second human opinion is needed to make a decision though.

7. Is also valid criticism. I think by the time I reached this test case, I was growing tired of doing everything as strictly as possible, so it's true that the quality of this one test was actually reduced as a result. If nothing more pressing comes up, I can address this one tomorrow morning. In the meantime it's also fine if someone else wants to step in.

I am still uncomfortable with on_missing. Why are things missing in the tests to begin with? We really need to understand rather than just bandaid over it.

Also is this still relevant or can this be simplified after #19740 ?

Why are things missing in the tests to begin with?

This is absolutely expected as a result of my intended changes: previously these functions silently relied on get_cache_dir creating the directory if it wasn't already there, so they were all written within the assumption that it couldn't be missing by the time they got a path back. I'm intentionally changing them so they don't force the directory to exist, so on_missing isn't a band aid, it's an explicit way to express who's responsible for checking and/or ensuring the dir exists if we actually need it at all.

Also is this still relevant or can this be simplified after #19740 ?

It's still relevant. I think these are actually 2 different problems.

Point 5 is actually nonesense

It's possible that I misread the description. This should be double checked

Let's check/do 5, 6, 7 in a follow-up

this patch was only meant for 8.0+, it cannot (nor needs to) be backported to 7.2.x