It turns out that we had not been turning off the PRCL FF properly (likely because it was added after MICH and SRCL FF, and I forgot the turn off step in the DOWN state....ooops).  However, we had incidentally been turning it off by clearing the output matrix when preparing to lock ALS DIFF, in the ALS_DIFF guardian. 

I have now added to the ISC_LOCK DOWN state to turn the PRCL FF gain to zero in the same area that the MICH and SRCL feed forwards' gains are set to zero.

In the end, this has not been hurting our ability to relock too much, since the actuation to the ETMs from PRCL was shut off during most of the acquisition sequence. We were still sending signal to the ETMs while trying to lock the green arms, but with the PRM misaligned during that state it should have been nearly zero, so probably hasn't been hurting too much, I hope.