False positive: `TypeVar`s in functional-style`TypedDicts` are allowed but unsupported

Describe the bug
Pyright currently allows users to define a TypedDict with functional syntax that contains pre-PEP-695 TypeVars; however, the resulting type cannot be subscripted.

I believe the correct behaviour would be to emit an error at the definition location, informing the user that TypeVars in that context are meaningless.

On a side note, is it currently possible to define a TypedDict that is generic over the type of a key that would not be a valid Python identifier?

Code or Screenshots

from typing import TypedDict, Generic, TypeVar

_T = TypeVar("_T")

# --- class syntax --- #

class Foo(TypedDict, Generic[_T]):   # Generic class-style TypedDicts are supported
    key: _T

def foo(x: Foo[int]):
    reveal_type(x["key"])   # 'int'

# --- functional syntax --- #

Bar = TypedDict("Bar", { "@key": _T })   # no error here

def bar(x: Bar[int]):       # error: 'Bar' does not accept type arguments
    reveal_type(x["@key"])  # '_T@Bar'

(pyright playground)

VS Code extension or command-line
I encountered this problem with the basedpyright VS Code extension, but I can confirm the pyright playground (v1.1.407) exibits the same behaviour.

Start by reproducing the functional-style TypedDict example in the linked Pyright playground, focusing on the accepted _T in TypedDict("Bar", {"@key": _T}). Trace the functional TypedDict validation and generic-subscript handling. Done means the definition reports that the TypeVar is unsupported or meaningless, with behavior covered by a regression test; the question about non-identifier keys may need separate investigation.