flake check too slow; causing too many PRs to fail. (#2342)

We've seen endless PRs fail due to this action taking too long or using
up too much disk space on the GitHub runners.

Until we can find a way for this check to be consistently cached or
otherwise evaluated consistently (i.e. by having it run on our runner),
we just have to drop it.

See also - #2325

Author: noonio