[red-knot] Support metaclasses

[AlexWaygood]

Currently we infer that all class objects are instances of builtins.type:

ruff/crates/red_knot_python_semantic/src/types.rs

Lines 1211 to 1212 in b7e32b0

	// TODO not accurate if there's a custom metaclass...
	Type::ClassLiteral(_) => KnownClass::Type.to_class(db),

Contrary to the TODO comment there (not sure who wrote that), this is accurate -- all classes are indeed instances of builtins.type. However, it could be a lot more precise, since some classes have a custom metaclass!

>>> class Meta(type): pass
... 
>>> class Foo(metaclass=Meta): pass
... 
>>> type(Foo)
<class '__main__.Meta'>
>>> isinstance(Foo, Meta)
True

For the Foo class here, Type::to_meta_type should return a type representing the class object Meta rather than a type representing the class object builtins.type.

---

Miscellaneous details on semantics

If a class is a subclass of a class with a custom metaclass, then the subclass will also have that metaclass:

>>> class Bar(Foo): pass
... 
>>> type(Bar)
<class '__main__.Meta'>

And if the metaclasses of the bases conflict, then this can happen (we'll need to detect this and emit a diagnostic):

>>> class A(type): ...
... 
>>> class B(type): ...
... 
>>> class C(metaclass=A): ...
... 
>>> class D(metaclass=B): ...
... 
>>> class E(C, D): ...
... 
Traceback (most recent call last):
  File "<python-input-4>", line 1, in <module>
    class E(C, D): ...
TypeError: metaclass conflict: the metaclass of a derived class must be a (non-strict) subclass of the metaclasses of all its bases

We'll probably need to add a method to ClassType to retrieve the type of the metaclass. It will probably be best to wait until #14087 has landed before working on this.

[AlexWaygood]

Could be a good @charliermarsh -sized task!