Correcting RoI part of HLTCaloCellMaker and fixing bug in TrigCaloDataAccessSvc

New merge request to improve RoI part of the HLTCaloCellMaker algorithm, which was only dealing with one layer (was actually in a prototype mode). Also, when checking regions not supposed to be covered by a particular layer, the data preparation service (TrigCaloDataAccessSvc) needed a fix to handle such cases. Should be a simple change, but certainly should give difference in tests.