introduce 'randomness' k/w to mark tests which may vary from run to run due to usage of (pseudo)random. it's important to mention that the same k/w is used w/ the exact same meaning in the jdk test suite, so this will also further reduce differences b/w these two test bases.

this is just a preparation state, the actual marking will be done by other sub-tasks of JDK-8241623 (most probably component-by-component).