Optics tests are taking too long to run. Email from Peter:

The delay is when the particles are being fired. Sometimes it zips through all 13 test particles in about a second. Sometimes it takes about a second per fired particle. I don't understand why. I could probably bring the time down on one of the classes, but the PolynomialOpticsModel has to create polynomial maps to be tested properly, and the way it does that is to fire test particles as you know. Hmm...I suppose what I could do is put an option to use canned data to build the optics model. It's a bit hacky but it would compensate for the problem. I'd much rather understand why the time it takes to fire particles is so inconsistent.