Fixes #14098 

Having invalid statements in a NamedTuple is almost like a syntax error,
we can remove them after giving an error (without further analysis).
This PR does almost exactly the same as
#13963 did for TypedDicts.

Co-authored-by: Shantanu <[email protected]>

Ref #12840

Fixes #12483
Fixes python/typeshed#7564
Ref #12841

The fix is straightforward. I can't use a unit test for this because
there are some builtins fixtures that don't have tuple, so I can't do
version check.

- Added missing cast import
- Changed revealed type

Co-authored-by: Shantanu <[email protected]>

See #14130 for context.

Btw it looks like these `Any` reports are quite broken in general. Some
issues I found:
* Many types are reported twice (even non-recursive)
* Explicit `Any` in alias r.h.s are not counted (because of reckless
`res = make_any_non_explicit(res)` in semanal.py)
* For generic aliases we count their r.h.s. as containing `Any` from
omitted generics

I tried to fix these things, but it is not trivial, so maybe we can do
it later in a separate PR.

Fixes #14139

#14148)

Fixes #14137 

Fix is trivial, I just forgot to call `super()` in one of my previous
PRs.

While working on another PR I noticed that current behavior of
`has_recursive_types()` is inconsistent, it returns `False` is there is
a recursive type nested as an argument to a generic non-recursive alias.
I wasn't able to find any situation where this actually matters, but I
think it is better if this function behaves consistently.