L2SA resolution modeling and L2SA pT extrapolation to IP for muComb

There has been a long-standing issue in muComb due to wrong L2SA pT resolution modeling. We also found L2SA pT is wrongly extrapolated to IP in muComb, even though L2SA pT is already the value at IP. So energy loss in the calorimeter is counted two times. This MR updates the L2SA expected resolution, and fixes these issues, based on 2022 data. In |eta| ~ 0.6, we found poor L2SA pT resolution, so tentatively assign an expected pT resolution of 100% for the region. The efficiency curves below are estimated with Zmumu MC. They show better L2CB rejection of low pT, keeping plateau efficiency.

ping @okumura, @vippolit , @yochi , @sshaw and @tamartin