Difference between cph109 and cph10c

Problem from Ray:

Yanwen has compared the distribution of the number of uniabsed CPR clusters associated to each EmObject between 4.8.5 production and 5.1.1 production. In the later produciton the number of clusters seems to be increased by about 25%. I know Tanya fixed some logic concerning how overlapping clusters were handled but I did not think it made this much difference. Does this difference make sense? What might have caused it?
plot at: plot number 6 (misc), 5th histogram.

Answer from Tania:

I understand the reason for the difference and I think the current coding should be right.

Let me first tell the differences. In the previous code the checking for the used wires was in the cluster making stage but since for the track based wires we do not want to mark wires as used, I put the checking for used wires in the wire collection maker. This should not cause any problem. However, in version 4.8.4, a cluster had to have fixed N "unused" wires. So if any of the wires around the seed wire was used by another cluster, this cluster wont be formed.

In the current version, a cluster could have less than the required N wires, so its a flexible window and if one of the wires is found already used in another cluster, that wire is dropped but the cluster is still formed with the remaining wires.

So in cases of larger occupancy events where we have overlapping hits, we would expect to see some differences.