Meaning of Statistical Uncertainty in PAD

I'm trying to understand the statistical uncertainty entry in the CLs_output_summary.dat from a PAD analysis and would appreciate some clarification. Is the statistical uncertainty the uncertainty in the maximum signal cross section allowed for that signal region, in pb? Or the same uncertainty express as a fraction of smax so that smax*(statistical uncertainty) is the statistical uncertainty in smax? Or the uncertainty on some other related quantity like the maximum number of signal events?

Also, could you clarify if the statistical uncertainty represents the uncertainty in the limited number of signal events expected in the signal region for that analysis (i.e. cross section*efficiency*luminosity), or uncertainty from the sample size of MC simulated events that passed the cuts in that SR, or some combination of both?

Lastly, to double check, the efficiency in the same line of the CLs_output_summary.dat is the fraction of events that pass the SR cuts for unweighted events or fraction of weights for weighted events, correct?

The statistical uncertainty in the file isn't quite matching my expectations, so I suspect I'm misinterpreting some value in CLs_output_summary.dat.