Allow for simple definition of (approximate) background error to be used for estimated significances in cutflows and subplots

For many initial optimization studies it may not be sufficient to determine a cut&count significance estimate taking only Poisson statistics into account but one may want to include some basic (background) systematic uncertainty estimates.

Therefore it should be made possible to specify a (list of) paths with should be taken as a systematic uncertainty. The scaling of the uncertainty (e.g. 50% of some process) would be performed using the existing path artihmetics.

Example usage (via visualize master config):

significance.bkgErrorFromPath: 0.5*bkg/$(channel)/$(campaign)/fakes/fakeE, 0.3*bkg/$(channel)/$(campaign)/fakes/fakeM

If (as in this example) multiple individual contributions are given (comma separated) the background error should be taken as the sum in quadrature of the individual contributions.

The same option should automatically be used by both, plotters and cutflow printing. It should, however, only be considered for the significance, not for individual yields shown (as matching of "error paths" to "central-value paths"/processes would be complex and possibly unreliable)