Fix doctest basenames when Sage library cannot be found with SAGE_LIB

[mkoeppe]

In fact, we remove the use of SAGE_LIB and replace what depended on this by recognizing certain package data directories of the Sage library.

• Based on a rebased version of doctest: Make sage.doctest.sources.get_basename work in more situations #22445 by @infinity0 and @saraedum.
• Part of Meta-ticket: Remove use of SAGE_LIB and SAGE_EXTCODE variables #33037.

Fixes doctest failures reported in #37645.

Fixes #22445.

📝 Checklist

• The title is concise and informative.
• The description explains in detail what this PR is about.
• I have linked a relevant issue or discussion.
• I have created tests covering the changes.
• I have updated the documentation accordingly.

⌛ Dependencies

[mkoeppe]

@tornaria Please take a look if this is solving the problem for you. (I haven't tested very thoroughly yet.)

[kiwifb]

I will have to look at this carefully.

[tornaria]

@tornaria Please take a look if this is solving the problem for you. (I haven't tested very thoroughly yet.)

No, it doesn't. As a matter of fact in my example from #37645 sage.doctest.sources.get_basename gives the correct output already without this PR. It's only when running the test that it prints just the name of the package without qualification (i.e. tolerance instead of sage...).

Thus, this PR is not fixing anything broken.

[mkoeppe]

As a matter of fact in my example from #37645 sage.doctest.sources.get_basename gives the correct output already without this PR. It's only when running the test that it prints just the name of the package without qualification (i.e. tolerance instead of sage...).

That's rather surprising. If you have a moment, could you share how you tested this?

[tornaria]

As a matter of fact in my example from #37645 sage.doctest.sources.get_basename gives the correct output already without this PR. It's only when running the test that it prints just the name of the package without qualification (i.e. tolerance instead of sage...).

That's rather surprising. If you have a moment, could you share how you tested this?

The link is up there in the description of #37645: #37138 (comment)

I did exactly what I describe in that comment (which passes all tests in normal mode) and then repeat the last line but with --long (the failing tests in sage.doctest.test are all # long time).

[github-actions]

Documentation preview for this PR (built with commit 2f81d78; changes) is ready! 🎉
This preview will update shortly after each push to this PR.