As reported in the TGM - https://indico.cern.ch/event/1129558/#8-l1-simulation-update-l1-calo a fix to the eFEX inefficiency at |eta|>0.8 has been found. The issue was a mismatch in the numbering scheme that is used to simulate the different eFEXes.
LVL1RoIs_output_fixedFresh.pdf - slides 20 and 23 show that the fix changes only the outer eta regions of the egamma and tau algorithms, but not the et distributions or phi locations of the TOBs.
This merge request is urgently needed for the next round of validation. @dzanzi @tamartin @aranzazu @okumura @vmartin