fix phiBE/etaBE for clusters in crack with negative energy in some samples
Fix phiBE/etaBE for clusters in both barrel and endcap, when one (only one) of the sample energies is negative
This could get rid of weird phi (outside -pi,pi) for clusters, as observed, e.g. here https://its.cern.ch/jira/browse/ATLASG-2486
Edited by Jean-Baptiste De Vivie De Regie