Use import_module instead of find_spec

[user202729]

Otherwise the test would fail with meson editable install. See https://github.com/sagemath/sage/actions/runs/13003203795/job/36265539648 .

Looks like the function was last changed in #36407. There was no discussion why the simple implementation is not used.

This is part of the fix for this test. The other part needed is #39498

📝 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. (can't really test, but see Test pip editable install with meson #39369)
• I have updated the documentation and checked the documentation preview. (no documentation change)

⌛ Dependencies

[github-actions]

Documentation preview for this PR (built with commit ef72c43; changes) is ready! 🎉
This preview will update shortly after each push to this PR.

[tobiasdiez]

LGTM. @tornaria is this change fine with you?

[tornaria]

LGTM. @tornaria is this change fine with you?

I've no idea, this code is very old. I didn't write it, only fixed it to work on python 3.12.

The original code comes from from #15351 (comment). AFAICT, nothing uses this function load_submodules() so it should be irrelevant what you do about it, even remove it, but I don't know its purpose, maybe @videlec can comment.

I'm not using meson to build sagelib. For my void package I use the sagemath-standard sdist from pypi, which I don't think allows building with meson. Is there a published sdist to build sagelib with meson?

[tobiasdiez]

@tornaria Thanks for the feedback. Tests are green (apart from the usual random failures), so let's get this in.