Change LAr sample sums & sample square sums used by autocorrelation to 64-bit integer

Intermediate objects used in autocorrelation calculation (sample sums and sample square sums) were exceeding the size of the 32-bit integers that they were assigned in some cases for the digital trigger, where pedestal values may be higher. All instances of these objects were therefore changed to 64-bit integers. Some other fixes to job options are included as well (e.g. a bug fix in LArExample/LArCalibProcessing/share/LArCalib_Ramp_jobOptions.py)