warn when default values on overloads differ from implementation signature

from typing import overload


@overload
def foo(value1: str, value2: bool = False) -> str: # this default value is wrong/useless
    ...


@overload
def foo(value1: int, value2: bool = False) -> int: # this default value is wrong/useless
    ...


def foo(value1: str | int, value2: bool = True) -> str | int:
    ...

in fact, maybe even prohibit defaults on overload signatures entirely, like in typescript:

function foo(value1: number, value2: boolean = false): number //error: A parameter initializer is only allowed in a function or constructor implementation.
function foo(value1: string, value2: boolean = false): string //error: A parameter initializer is only allowed in a function or constructor implementation.
function foo(value1: number | string, value2: boolean = true): number | string {
    return 1
}

Start by reproducing the overload example and trace mypy's overload-signature checking. Determine whether the accepted behavior should be a warning for defaults that differ from the implementation or an error for any defaults on overloads. Done means the chosen rule is enforced consistently and covered by regression tests.