Propagate scale uncertainties to histograms

This patch introduces the ability to include scale uncertainties in histogram total background uncertainty bands. This is accomplished by retrieving the counter for /bkg and its relative scale uncertainty, then applying this relative uncertainty to every bin.

In this current iteration, this is done automatically if experimental systematics are retrieved. It could probably use some fleshing out, even if it's just adding an explicit option to retrieve these scale uncertainties, but I wanted to get some feedback on whether it would ever be desirable not to propagate the scale uncertainties to histograms.

I should also add that the changes to TQSampleDataReader are just to avoid a segfault when specifying readAnalysis.NF.toySize: 1, since the code will otherwise go looking for correlations and try to access a folder that isn't present. I hope it doesn't cause any bad behavior for those using correlated uncertainties, but I haven't a way to test this...