Update ty to 0.0.86

[srittau]

Ignore rule invalid-property-type-override with the same reasoning as the other incompatible override rules.

[srittau]

Cf. #16551

@JelleZijlstra @AlexWaygood Maybe one of you could have a look?

Claude also tells me:

One looks like a ty bug. In antlr4/Parser.pyi:80, state: int overrides Recognizer.state, a property whose setter takes int. ty says the override "does not accept writes of type int", but a plain int attribute obviously does.

I agree with Claude here. Even if overriding a property with an attribute is not accepted (although I'd argue that it should be in this case), at least the error message is misleading. Should I report this to ty?

[JelleZijlstra]

That does sound like a bug, but I can't currently reproduce it: https://play.ty.dev/229cf4e3-8fbe-4d65-91ec-34233f065837 . Either it's already fixed or there is some aspect of the stubs that make it behave differently.

[JelleZijlstra]

Codex figured out the missing ingredient: these classes have slots. Reported astral-sh/ty#4745.

For typeshed, we can also probably remove the state: int declaration in the subclass, as it's redundant.