Earlier the squeezer unlocked and took us out of observing. 

• The SHG guardian had a check on the transmitted power that makes it think that the SHG is unlocked when the normalize transmitted power drops below 0.7. You can see in the attached 3 day trend the SHG power gets close to this level occasionally, so for now I've lowered the threshold to 0.6.  Once this change was made the squeezer relocked and we were able to go back to observing before the wind knocked us out.
• Ed tried the procedure to have us be in observing without squeezing, 48267.  The problem we encountered was that if we don't loose lock bu the squeezer state changes, ISC_LOCK recognizes that the squeezer state has changed and complains about that.  We attempted to manual back to INJECT_squeezing, so that ISC_LOCK would make the right request, but that still didn't work.  The problem might have been that SQZ_MANAGER wasn't OK, even though it was in the nominal (NO_SQUEEZING) state.

Changes to SQZ_MANAGER:

• The squeezer manager wasn't managing the SHG guardian, and had no check that the SHG was locked.  It recognized that the OPO was unlocked, but it was sitting in PREP_IFO_SQZ while te SHG was unlocked.   I find this situation confusing, and think that others will as well, becuase the squeezer wasn't ready for injecting into the IFO.  I've made several changes to the SQZ_MANAGER guardian, which I think will make things more transparent.  You can compare the old and new graphs attached.  There is now a state for locking the SHG, and PREP_IFO_SQZ has been separated into three states, LOCKING_OPO_DUAL_RESONANCE, LOCKING_CLF, and SQZ_READY_IFO.
• I also added decorators to appropriate states that check for the SHG lock and the OPO lock.
• I renamed IDLE to LOCK_TTFSS, which is what that state was checking for, and moved where it was in the graph.  (In the old graph there are many arrows that point to the IDLE state, but there were no circumstances where these paths would be used so I've deleted them.
• I split the state which is was called SQUEEZING into two states, SQZ_ASC and SQUEEZING, just to make it more clear to the user what is going on in each state.  
• There are a couple more changes I'd like to make that I didn't have time to do and check while we were relocking:
  • Move the management of the beam diverter into the manager, out of the OPO guardian.  
  • I would like to have a closer look at some of the error checking (for example in the SQUEEZING state). 
• The way that ISC_LOCK was requesting squeezing to be injected, it would not wait for SQZ_MANAGER to arrive before it moved on to low noise, I edited this so that we can't jump to observing while the squeezer ASC is still coming on, but haven't loaded the change since we are now in observe.