Regression in definition of `UUID.int`

[erictraut]

A recent change in uuid.pyi results in a regression when accessing the int attribute of the UUID class. See this pyright bug for details.

At runtime, it is an error to have a __slots__ entry share the same name as a class variable. The UUID class definition violates this invariant.

To fix this, I think the UUID class definition should be modified in one of two ways:

1. Remove the int from the slots definition, reverting the recent change
2. Remove the int property and add a self.int: builtins.int instance variable annotation within the __init__ method

Option 2 is probably better since it more accurately models the runtime.

[srittau]

At runtime, changing UUID attributes is prevented via __setattr__. This is modeled using @property here, and would be better modeled using PEP 767's ReadOnly. For now, I agree that it's better to change int to a regular instance variable. We could also experiment with it being Final, even if that's not really true at runtime. Finally, we should also __setattr__ returning NoReturn, since that's what's happening at runtime.