let the category of sets handle the __len__ method of rings

[fchapoton]

This removes the __len__ method in ring.pyx and let the category of infinite sets handle this.

The test with RR fails in some other bizarre way, so it is marked as a bug.

But maybe it would be better to just remove __len__ from rings instead ? People should use cardinality.

📝 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 and checked the documentation preview.

[github-actions]

Documentation preview for this PR (built with commit 23653fb; changes) is ready! 🎉
This preview will update shortly after each push to this PR.

[tscrim]

The issue with RR is that it is an extension class and __len__ is a special slot method in Cython. You can also see this failure with ZZ. So the thing to do is to change the __len__ implementation to be in parallel to Parent.__pow__.

[fchapoton]

thanks a lot, Travis, for your continuous help and useful insights !

[tscrim]

No problem. Thank you for all your work on improving Sage!

[tscrim]

You still have the issue with RR, ZZ, etc. correct?

[mantepse]

I agree that __len__ should not be a method in the category of rings. What is the reason for having it?

[fchapoton]

Should I use "NotImplementedError: infinite set" or "ValueError: infinite set" ?
Currently, both do appear. It may be a good occasion to have more uniformity.

[tscrim]

I would say it would depend on if we expect the infinite set version to (eventually) work/be-implemented or not. Granted, we aren't consistent about this either... I would lean to ValueError as a general rule.

[fchapoton]

So should I change the existing "NotImplementedError" ?

[tscrim]

If you want to, go ahead.

[fchapoton]

Well, I finally would prefer not to do further changes here.