Many tests are broken due to the newly added note, I'd like to first confirm the wording and availability of this PR before pervasively modifying the tests

I almost forget about this PR.....Now I modify the test suite to apply the new change and pass the tests. But I am worried about that this PR may be too pervasive

I was just pointed to the docs introduced by this PR (Variables vs type aliases) and for a few moments I was very confused.

Perhaps this was because I was trying to create a type alias for Annotated, which apparently doesn't work (though I think it ought to).

But the docs (copied verbatim from Ivan's text in #3494) did not help. I think what the docs are trying to point out is that both a variable with type Type[A] and an unannotated "variable" initialized with A hold the class object A at runtime, but the type checker sees them differently: the variable tp is a variable holding a class object (that must subclass the given class), and (being a variable) cannot be used in annotations; while the alias is a type definition to the type checker, usable in annotations, and (like other type definitions) is not allowed to be changed.

I'm still not sure how to best write this down, but I feel that the docs as they currently exist could still use some improvement.

@gvanrossum Thanks for the feedback! The docs here definitely needs refinement since #9238 also points out its incompleteness. I'll file a PR to update that and will request for your review there.