Optimize implementation of TypedDict types for **kwds

The implementation copied lots of callable types even when not using
the new feature, which was expensive. Now we only generate a copy if a
callable actually uses TypedDict types for **kwds.

This made self check 7-8% faster (when compiled with -O0).

The original implementation was in #13471.

Author: JukkaL

mypy/types.py

@@ -1976,7 +1976,7 @@ def expand_param_spec(
 
     def with_unpacked_kwargs(self) -> NormalizedCallableType:
         if not self.unpack_kwargs:
-            return NormalizedCallableType(self.copy_modified())
+            return cast(NormalizedCallableType, self)
         last_type = get_proper_type(self.arg_types[-1])
         assert isinstance(last_type, TypedDictType)
         extra_kinds = [
@@ -2126,7 +2126,9 @@ def get_name(self) -> str | None:
         return self._items[0].name
 
     def with_unpacked_kwargs(self) -> Overloaded:
-        return Overloaded([i.with_unpacked_kwargs() for i in self.items])
+        if any(i.unpack_kwargs for i in self.items):
+            return Overloaded([i.with_unpacked_kwargs() for i in self.items])
+        return self
 
     def accept(self, visitor: TypeVisitor[T]) -> T:
         return visitor.visit_overloaded(self)