[BUG] Intermittent failure of download in testing workflow

[teald]

Describe the bug
During some test workflows (e.g., this workflow run), astrodata.testing.download_from_archive will get a timeout error while attempting to read in a file.

To Reproduce
Unknown, it's unclear if this is a problem with astrodata, GitHub Actions/the workflow file, or something external.

Expected behavior
Downloads should complete without timing out.

Example code
See e.g., workflow run

Desktop (please complete the following information):

• OS: variable (occurs on ubuntu, macos, windows intermittently)

Additional context
It seems like recent workflow runs may be working again, but it's unclear why.

[teald]

This (external) issue has been resolved.

[teald]

The macos runners are having problems related to this issue.

[teald]

This has been resolved. Also, IP addresses are now reported in the testing runs (see PR #34)

[teald]

This issue is ongoing during some service outages with the gemini archive.

[teald]

This is frequent enough---and will be for a bit I suspect---that it's probably better to have some of the failing downloads xfail and have that escalate in special cases (e.g., it must pass before a release, or a PR). Usually one/two re-tries will succeed without issue.

If there's a straightforward way to have the action retry itself, even better.

[teald]

Updates to the archive should have this solved 🤞