Implement energyErr systematic variation

I implemented a systematic connected to the non-closure of the sigma_m/m variable. We base this on the non-closure of the energyErr/E (or energyErr, as we do not change the energy for this systematic). In the full Run 2 measurement, initially 5% was used, but it was reduced to 2%: https://indico.cern.ch/event/1131803/contributions/4758593/attachments/2398621/4111806/Hgg_Differentials_Approval_080322.pdf#page=47

I followed this approach with the initial 5% uncertainty, which can be reduced later. We should check the agreement in Zee and m(yy) (base) sidebands to assess if it can be reduced.