Open
Description
Remark (for myself): improve this diagnostics when I get to untangling the runtest logic to make it so that test suites explicitly declare their intended default behavior so that the diagnostics here can say e.g. "expected to check-fail, but this didn't fail" or whatever.
Originally posted by @jieyouxu in #133813 (comment)
E.g. ui
tests by default are expected to "check-fail", compare that to the actual test outcome, etc.