Investigate the Harmonic mean p-value procedure

It is possible that the Harmonic mean p-value procedure for deciding pass/warn/fail for groups of test results might be better than just a global p-value correction based on the number of tests. This should be investigated, and probably tried out.