Allow using Unpack with arbritary bounded TypeVars

[PythonCoderAS]

I want to make a class where the values of kwargs are specified through a TypeVar, something like this:

from typing import *
from abc import ABC

class EmptyTypedDict(TypedDict):
    pass

DictT = TypeVar("DictT", bound=TypedDict, default=EmptyTypedDict)

class MyClass(ABC, Generic[DictT]):
    @abstractmethod
    def my_func_inner(self, **extra: Unpack[DictT]): ...
    def my_func(self, arg: int, **extra: Unpack[DictT]):
        print(f"Arg: {arg}, extra: {extra}")
        return self.my_func_inner(self, **extra)

class SubExtra(TypedDict):
    x: int
    y: str

class MyClassSub(MyClass[SubExtra]):
    def my_func_inner(self, *, x: int, y: str):
        print(f"x = {x}, y = {y}")

Unfortunately, there seems to be no way to do this, firstly because TypedDict is not a valid bound type, and second because Unpack will not take TypeVars.

I'm thinking that maybe we could make it so that TypedDict can be a proper bound, or an equivalent bound type that will make Unpack happy.

[brianschubert]

There’s an open feature request for just this: #1399