Improve how PyCBC Live deals with bad data in frame files?

[titodalcanton]

Low-latency frame files might in rare occasions stop containing one of the required channels (strain data or state vector). In that case, PyCBC Live's current frame reading code will cause it to reach the timeout limit, behaving effectively as if the frame files did not exist. Although this is not necessarily a bad behavior, maybe it would be better to immediately treat the interferometer with missing channel as not analyzable, so as not to unnecessarily slow down the analysis of the other interferometers.

While looking into this, and following various discussions in the LVK, I also realized that one could first check the state vector, and if the state vector indicates analyzable data, then try to read the strain channel. PyCBC Live currently does the opposite, i.e. read the strain first, then the state vector, and then drop the data if the state is bad. I am not sure how much practical advantage changing this would bring, but at least it feels like a more correct approach, and cleanly allows frame files to drop the strain channel when the detector is not observing.