Improve record_assert_trip mechanism
According to discussion here: #323 (comment 98642) we should:
- unify message logging (probably best to put both messages to wlog)
- improve recorded message to additionally contain also full message produced by assertion (now it includes only condition)
- try to implement explicit cancellation of recorded trip, to avoid records specific to cases when assert throw was really ineffective, like in case described in issue #323 (closed)