This algorithm can be scheduled after clusterization and monitors some variable (ends up in histograms and tuple). Local vars for now but trivial to extend. This can also be used directly on an AOD file provided the clusters have been persistified.
/cc @dntounis @aleopold