fix rare FPE in CaloScaleCluster.cxx
One event in the million-events MT test ran on Jan 1st showed an FPE in the LArClusterMaker.rfac_v5_ele35
correction tool. For one particular cluster in one event, the correction factor winds up to be 0.0. The cluster in question has exactly eta=0 and phi=0.
This MR adds a protection to avoid the div-by-zero.