While we were out of observing I did a few quick things to try to make our locklosses a little gentler.  (part of WP 6391)

First, I set limits on the TR QPD pit and yaw signals of 2, rather than 100 which were the previous limits.  These shouldn't go above 1 if we are locked, but can become large after a lockloss and send large signals to the suspensions through the soft loops.  I accepted the new values in SDF so we could go back to observe.

I did a little bit of work on the down state of the ISC_LOCK guardian, and loaded some of the changes.  

• One problem was that it was setting TRAMPS, GAIN and then clearing history for the same filter bank in consecutive lines, which we know doesn't necessarily happen in order, so this might set the gain before setting the tramp, and certaintly will clear the history before ramping the gain down.  
• There were some things that were done in both the ISC_LOCK guardian and in ISC_DRMI, (handling of DRMI ASC) which I've removed from ISC_LOCK.  
• moved the things that are importatnt to the begining of the state, so they will happen faster.  

We can see if this makes things better for the next lockloss.