Rethink VBF HH treatment in LO hh-limits

[tolange]

Currently we include the VBF SM hh sample in all LO limit scenarios (as other analysis do as well) This is not completely correct but an approximation. Currently this is not done properly however, since the LO ggf samples are scaled to 1 pb while the VBF sample is scaled to a proper XS. We need to either remove VBF here or we need to downscale the ggf samples to a proper XS (or upscale the vbf samples, but to which XS? 1/sigmavbfSM * sigmavbfsm/sigmaggggfsm ?)