The strict/tolerant mode mainly works by using either a strict or a tolerant regex.  If the markup is invalid, the strict regex doesn't match and it gives an error.  The tolerant regex will match both valid and invalid markup at the same time, without distinctions, and that's why there's no way to emit a warning for these cases.  I think there are a couple of places where a warning could be emitted, but that would just cover a small percentage of the errors.  Even if we find a way to emit a warning for everything allowed by the tolerant mode that fails on strict, it won't still cover all the possible errors, that's why I think tools like validators and conformance checkers (or even the warning/error logs) should be used instead.