Fix: switching from eta to absolute eta in METSignificanceTool
This MR fixes a bug discovered by @hpacey: make the handling of pile-up uncertainties in the METSignificanceTool actually eta-symmetric by switching from eta to the absolute value of eta.
Tagging also @mswiatlo and @camacho.
We'd also like to sweep this into master, but I'm not sure which label I've to set for this. Could one of the shifters maybe help with that?
Thanks!
Michael