Calo{E,S,L}Corrections: Prefer LHCb::Math math functions over local implementation

as this will select the right implementation for both double and float, and whill thus make it easier to try and switch to float (in a subsequent MR)