This is the distribution of phi with the current version of analysiscut2:
Here is the distribution from "Other Possible" showers found by Chao's code (seems to go from -90 to +270):
If I change the parameter limits of phi from to -pi...pi, and this seemed to help the distribution. Still some showers stuck at +/- 180:
I think I will keep this version.
Also added change to analysiscut2 to ignore any hits outside 20us. These are almost never part of the shower.