Implements basic method introspection

[Tpt]

Implementation notes:

• I am not sure about the decorators for class attributes. I am using the @classmethod\n@property combination that got removed in Python 3.11. I don't see a clean way to do it in the next Python versions. It's possible that type checkers work properly with it, we will need to check but I would prefer to focus on other things for now.
• When we introspect the class definition (#[pyclass] macro) we don't know the list of method so each method is generating its own introspection fragment with a "parent" attribute that connects a method to its class. We will likely reuse it again for nested classes used in complex enums.
• We are generating a / after the self or cls argument even (if they are not other positional-only arguments). It's not nice but syntactically correct.
• I have not tested all magic methods

[bschoenmaeckers]

I am not sure about the decorators for class attributes. I am using the @classmethod\n@property combination that got removed in Python 3.11. I don't see a clean way to do it in the next Python versions. It's possible that type checkers work properly with it, we will need to check but I would prefer to focus on other things for now.

Should we run a type-checker in the CI to verify the generated types?

[Tpt]

Should we run a type-checker in the CI to verify the generated types?

Yes, I am planning to do it when the implementation will be a bit more feature complete. At this stage it will complain a lot about missing elements in the stubs.

[davidhewitt]

Thanks, this looks good to me. Same as #5025, sorry I have been a bit slow here. I have just a few comments.

I am not sure about the decorators for class attributes.

I suspect the solution lies in either typing.Final or typing.ClassVar.

[davidhewitt]

Let's send this to merge, I guess it might conflict with #5089 👍

I lost track, do we have an issue listing all the future refinements to the introspection code? It would be helpful to record all the pieces which might be improved, both soon and once Rust supports them.

[Tpt]

Let's send this to merge, I guess it might conflict with #5089 👍

Thank you! Yes! It will conflict.

I lost track, do we have an issue listing all the future refinements to the introspection code? It would be helpful to record all the pieces which might be improved, both soon and once Rust supports them.

Not yet, I am going to do it

[Tpt]

Done: #5137

[Tpt]

@davidhewitt I have pushed a fix for the CI in this branch. The branches are conflicting, what about merging this one first and the next one after I do a rebase?