Removing phase 1 application on surface motion knm matrices. For negative knm calculations the phase was being revereds on the motion to optic knm but not to the optic to motion matrices. This led to unexpected sign errors in the PI gain. Originally thought the phase 1 application also needed to be applied to the surface motion knms, as it is applied to the static knms. On further thought this shouldn't be the case. Phase 1 is applied to make tuning with maps easier, this doesn't apply for a sinusoidally varying surface motion.