I honestly don't know; judging by the broken error messages, something must have gone horribly wrong in the trait resolver. I wouldn't trust any unsafe code to be sound.
And is this something that the compiler-team would like to be aware of?
Almost certainly. I've been thinking of locating the source of the error before filing an issue. (Maybe not today, though, since this issue has already broken my brain.)