Reports mismatches between the safety of a trait's definition and its implementation.