Thread history

Not sure what I broke in version 11.1. I'll be looking into it over the next few days. The changes I made were mostly cleaning up the code, so performance shouldn't have changed much.

Just to close off this thread, the problem with version 11.1 was with the wave cache I created for the 11.x series of robots. Wave creation is far more expensive with Precise MEA turned on in my Xander framework. At the same time, each guess factor gun would create a wave for itself when aiming before any actual wave yet existed. This resulted in pretty much the same wave getting created multiple times, resulting in the same Precise MEA getting predicted multipled times, which was absurdly expensive. The wave cache was meant to help fix the problem by allowing waves to be reused. However, the implementation was a bit flawed, and waves that were not quite identical could get reused. Ultimately, I really only needed to cache the Precise MEAs, so I changed it to an MEA cache instead of a wave cache, which actually simplifies it, and more importantly, fixes the problem.

One last note -- my Precise MEA calculation could be improved, and I may work on it in the future. The current implementation works, but it uses a shotgun brute force approach which I should be able to improve if I put my mind to it. What makes it so tricky is handling situations where the robot is near walls in combination with taking drive limitations into account (turn rate, accel/decel rates).