Add a method to allow certain system tests to be re-run, if they were marked as having failed in bin/tests/system/systests.output. This allows us to determine whether a test failure was transient or intermittent without having to run the entire system test job over again.
Ideally this should be run automatically from runall.sh or make test if there were any test failures in the initial test run, and it should provide an amended test summary that explicitly lists tests which failed but passed on retry, so that instability is clearly visible in test reports.