New Tag request: IMC_LOCKLOSS

At LHO we are recently seeing many locklosses where the IMC is loosing lock at the same time as the main IFO (in a normal lockloss it looses lock ~300ms later). See alog https://alog.ligo-wa.caltech.edu/aLOG/index.php?callRep=80366 from a plot and description of this. A similar thing happened in the second half of O3b but hadn't happened at all in O4 until September.

Iain Morton's "thresh" branch had some working code for this, mainly reused from !98. This code's refinement looks looks for crossing a standard deviation above or below the mean but thinking about it again, I don't think that's really needed for this tag in particular. See plot where AS_A dropped down and then spikes and would cross the above threshold 50-75ms later. If we improved this refinement it would bring the AS_A lockloss earlier.

Main things that need to be implemented:

• If the refinement is using a state higher than using the IMC channel to refine the lockloss (or some other ISC_LOCK state we think approipate), it should also calculate a refined time for the IMC loosing lock.
• We then need to a tag `IMC_LOCKLOSS` or some similar better name if the IMC refined time is less than 50ms AFTER the AS_A refined time. 50ms is what Iain used and it worked well...
• It might also be nice to add the IMC trend onto the refine plots. Maybe we'd need to zoom in more to see the difference...

LLO uses IMC to refine all their lock looses so this code change wouldn't affect them, though we should think about improving the LLO refinement.