Ok lets see if I can make this clearer.
You've stated that there are 2 types of collision checking:
1. Brute force.
2. Cell optimized.
If the debugger seperated those checks into 2 seperate counts then it would be easier to tell whether we need to alter our events because many of us use a lot of subevents (and you said that you may not have the cell optimized method if you don't use top level events).