[Corey, Jenne]

Corey was struggling with the MICH_BRIGHT step of initial alignment.  He let me know that the guardian kept looping through that set of states, and that MICH wouldn't lock.

Things in general looked fine (as in, the past things that have been struggles with this state), so I started looking at the MICH error signal.  It didn't look fine, even though the beam was reasonably well centered on the AS_A WFS diode.  There were PDH-looking jumps in the error signal, when we expect a MICH-only error signal to be a nice smooth signal.  That pointed me toward looking at the misalignment of various optics (PRM, SRM, ETMs), at which point I noticed that the OFFSET values for the TEST filter banks on the ETMs were zero rather than the ~10urad we normally have (one of the 4 was 0.5, but that's essentially zero for the purposes of intentional misalignment).  I realized afterward that this is why we were seeing IR flashes in the arm cavities according to the front wall TV - we shouldn't see those during MICH-only time, so we'll try to remember that that is a clue as to what is wrong.

These OFFSET values are also used by the new INCREASE_POWER INCREASE_FLASHES states that are helping to get the green arms locked when the optics are quite misaligned (this new state is probably related to why we haven't had to do very many initial alignments lately, which is awesome!).  Perhaps these OFFSET values weren't reverted after the testing work done yesterday with this state or they were left over from some other work or losing lock part way through the new state, so when the ALIGN_IFO guardian tried to misalign the ETMs it correctly turned on the OFFSET, but nothing really happened since the offsets were zero.

I reverted the values via SDF (SDF knew that the values were wrong, but it's hard to know which things that SDF says are differences are the things that are actually causing the problem...), and the initial alignment continued forward as normal.

Since this (incorrect offset values) is something that has caused us trouble a few times lately, TJ has been in contact with StuartA at LLO about modifying the suspension guardian to ensure that the offset values are set correctly every time we request the misaligned state.  We had decided to push this modification to January, but this is a good reminder that we should pick it up soon after the holidays, just in case this happens again.

EDIT: Corrected state name - thank you Corey!