segments.py mistakenly creates 2ifo jobs for 1ifo time

When a large stretch of time is analyzed at once, we break it up into more manageable segments. At present, these artificial boundaries can coincide with the time an IFO switches off. This leads to the creation of (for example) 2 IFO jobs in 1 IFO time; at the very least, this will cause reference psd jobs to persistently fail since it will try to write a 'None' time series to disk. We need to add some extra logic to prevent segmenting along existing segment boundaries in this module:

https://git.ligo.org/lscsoft/gstlal/-/blob/master/gstlal/python/segments.py

As a simple way to reproduce, filter over all of O1 at once and try to run the psd job that starts at gps time: 1127634250.