Regarding the discussion of tests expected to fail, and moving
expected-fail tests to different directories, I wonder whether you'd
prefer using an expected-fail annotation per test case.
An annotation seems useful for my own work, and lightweight, but I don't
know about core Racket (especially with the couple decades of legacy
tests done various ways).
https://www.neilvandyke.org/racket/overeasy/#%28part._.Expected_.Failures%29