BUG: fix StringIO uploads

[bsipocz]

This is a follow-up to #617 and does fix the issue I see in astroquery.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 82.31%. Comparing base (eba8c6f) to head (d5df951).
Report is 5 commits behind head on main.

Additional details and impacted files

@@           Coverage Diff           @@
##             main     #626   +/-   ##
=======================================
  Coverage   82.31%   82.31%           
=======================================
  Files          72       72           
  Lines        7429     7429           
=======================================
  Hits         6115     6115           
  Misses       1314     1314           

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

[msdemlei]

I get red lights flashing when I see strings and bytes in parallel. I have not tried to ascertain that something will be doing the encoding (to utf-8?), and I note that that is not necessarily trivial for XML (which as encoding indicators inside: a charset attribute in the xml declaration, defaulting to utf-8 if not present).

Given that, I'd really love to have some regression test for that. I'd even bite the bullet and have a remote test for it...

Or is there a simple way to see that the byte/str mix here is unproblematic in this particular context?

[bsipocz]

My take on this is that in case of streams we just pass along what we got already on the class instance. Whether that was provided by the user or went through some validation is a separate question.

That being said, I'm happy to add the tap upload test from astroquery as this one smoked the issue out: https://github.com/astropy/astroquery/blob/main/astroquery/alma/tests/test_alma_remote.py#L566

[andamian]

That remote test passes for both BytesIO and StringIO so I assume that requests encodes the string automatically. If requests users expect this to work, I see no reason for us to restrict it.

[msdemlei]

Ok, I'm happy with Adrian's explanation, and I trust requests will be doing something robust, so let's go.

[bsipocz]

I still meant to add more tests (practically moving or copying the one from astroquery), but it slipped off the radar.