Correctly initialise variables in the pufit calculation
There were some uninitialised variables in the pufit calculations that were probably behind the instability in the counts for triggers using these algorithms.
This should hopefully fix these (and ATR-20668)