After some pleading and finagling, I've got H1 locked again.  Attached is a screenshot with several DTTs, showing where this 4.2 Hz peak is appearing.  You can see that the ratio of the peak and the reference is largest in the SOFT loops, so perhaps it is getting coupled into there?  In the individual quad screen, you can see that EX has this peak much stronger than other quads, so as Sheila and Anamaria have proposed, it's likely a problem in DARM, but that is then showing up in the SOFT and then other angular DoFs. 

To get the IFO to lock, I asked Jim to wait for a long time at Power_30W for ADS to converge to the level of the noise floor, and then we also waited after Transition_from_ETMX before transitioning back to lownoise EX for the ADS to again converge to the lower noise floor. 

There were some funny changes in the Transition_from_ETMX state that I think were doomed to not work, so I've reverted all of them to how it was yesterday afternoon.  I suspect that these were why we lost lock in this state last lock.

Also, just in case the 4.2 Hz is coming from the DARM loop as a result of imperfectly matched analog and simulated L2 coil driver electronics, we've switched back to using the lowest noise L2 coil driver state.  So, I can't say if the reason I was able to lock is that I switched to the old coil driver state, or if it's because I have been waiting really, really long times for ADS to converge, but at least the combo of the 2 is working (with N=1 at least...).

Also of note, we have higher microseism today than we have been having lately, which is my suspicion for why we have worse noise at low frequency right now, which is showing up as a ~5+ hit to our BNS range.