As I was reworking the code to do the group statistics last night, I had one of those glimpses of where to go that you look for in a design. A few seconds of clarity when the Platonic ideal of what you're creating pokes through the messy aglomeration that is a working and evolving system.

Nothing huge, but there's some huge simplifications that can be made now that we're not tracking and updating 1/4 million triggers on every scan (which is a simplification from the specification that I've decided is the only reasonable way to go). There's some serious refactoring that would be necessary to make it work...

Refactor fearlessly indeed.