Systematics in applyMClumi

Currently, applyMClumi doesn't support systematics at all:

1. It cannot handle uncertainties in the cross section
2. It refuses to run on input files with gen-level systematics uncertainties

The code should support uncertainties in the cross section used to normalize background processes.

The framework should also be support for encoding generator level uncertainties with shape effects (e.g. PS weights, PDFs, scales...). One needs a way of specifying how to correlate them in the final fit (i.e. pass values for both v and i). Some of these weights would come from the Ntuplizer. In order to enable this, applyMClumi needs to support multiple gen weights as inputs.