Support namedtuple as one of multiple superclasses

[elazarg]

Solve issue #1558 by delegate subtype checks on TupleType to fallback.

This does not solve #1485. I don't know why.

[JukkaL]

Looks good to me, though there is one potential simplification. The if is_named_instance(right, 'builtins.object') check above your change now seems unnecessary, and maybe also the Sized check. Can you see if you can remove them to simplify the code a bit?

[elazarg]

builtins.object can be omitted. Sized cannot.

[elazarg]

Yes - more refactoring. It helped me understand what happens, and I believe it makes the logic clearer.

I can revert back to the minimal change if necessary.

[elazarg]

In the experimental refactor_to_enum branch I encountered what seems like another case of "Instance leak" we've discussed some time ago. I will try to investigate it; should I add the fix here or in a different PR?

[JukkaL]

If it seems unrelated to this change, it's probably better create a new PR.

[JukkaL]

The new methods don't really make the code easier for me to read. I'd suggest reverting to having them inline instead, as they are sufficiently simple.

[elazarg]

I would have opened an issue for the error, but it's hard to reproduce; in that branch there are many occurrences of the type, with similar context, and the error only occur in one of them.

The current patch fixes the error, but I don't think it addresses the root of the problem.

[JukkaL]

Okay if it's hard to isolate then no problem.

[JukkaL]

Thanks!