According to mypy_primer, this change doesn't affect type check results on a corpus of open source code. ✅

Just out of curiosity: Why is this check necessary? I just came across it when trying to update old TypeAlias uses to the new type statement. With this restriction, the latter is not really a replacement for the former, and I can't immediately see a semantic reason for disallowing it, and I couldn't find something in the PEP about this, either.

(Related issue: The new type aliases are not callable, but the old ones were. This is useful when e.g. aliasing a NewType. So again, the new type aliases do not seem to be a full replacement for the old ones.)