Fixes issues seen in ATLPHYSVAL-667 on changes in conversions with CL16

Default CaloEt used for back-tracking was set to 0. GeV for old CL in !34503 (merged), in the latest physval changes were seen in past CL16 where no changes were expected. The default is now sent to 4.5 GeV to match previous setups used in validations for rel22. The cut is still kept at 6.0 GeV at CL19 which is the default for track reconstruction and the output should be identical. Should fix https://its.cern.ch/jira/browse/ATLASRECTS-5645

FYI: @perjohan, @smwang, @nicolaid, @goblirsc, @sroe, @jferrand, @miochoa