As describe in the Trigger General Meeting (07/07), we are actually having cells with too high energy (or two low) only because the samples noise injected is being performed for 5 samples, instead of 7 (we need to keep one sample before, another after).
Warning @pavol too.
Relates to ATR-23178.