[Corey, Jenne]

We lost lock on the way up (not at all sure why, we hadn't yet gotten to the state we were worried about, but microseism is on an upward trend again), and when the IFO was relocking, the new power scaling (from alog 87806) did something funny.  

The guardian is set up to use the IM4 output if the IMC is locked, and the IMC input power otherwise.  It looks like this has worked successfully for the whole week since it was implemented, except for right now.

The MC2 Trans Sum was plenty high (but had only just gotten there), correctly indicating that the IMC was locked less than a second before the guardian had to decide whether the IMC was locked, so the Laser power guardian took for the power scaling the IM4 trans output.  However, it looks like the laser power guardian read that IM4 value, while IM4 trans was still settling, since it has some whitening / dewhitening.  

Once I realized that the power scaling was wonky, I just set the offset to 2, so that we could get back to locking while I investigated, which worked.  Since the laser power guardian does reset that value every time there's a change to the laser power, the guardian has already picked it up and put in the more correct value of 1.8.  

To prevent this from happening again, I put in a bit of logic to the set_power_scale() function of the LASER_PWR, such that if IM4 trans reads less than 1W, just use the input power as we used to do.  We basically never request 1 W or less from the PSL and also are locking the full IFO, so this should be fine.

I tested it by bringing the LASER_PWR guardian to 3W (it used the IM4 trans as it should), and then down to 1W (which ends up as about 0.9W at IM4 trans), and the guardian correctly used the input power for the power scale.  So, I think it should work fine.