improve jFEX coordinate treatment in L1TopoSimulation

Current treatment of jFEX coordinates obtained via globalEta() and globalPhi() function of relevant L1Calo TOBs have some subtle inconsistencies with the L1Topo firmware. This has been understood to be due to slightly simplified coordinates especially in the forward region where the L1Topo Firmware uses higher precision coordinates than the values returned by globalEta() and globalPhi() allow to infer.

With this MR the (exact) float coordinates from (decoded) jFEX TOBs are used instead and converted to the L1Topo internal integer representation analogously as it is already done for muons (which have an even more complex/fully irregular pattern requiring this way of treating them). As this would otherwise duplicate some logic the conversion from float to L1Topo internal coordinates has been moved out of the MuonInputProvider class into the L1TopoSimulationUtils such that it is re-usable.

Please note: this MR is expected to change some trigger counts slightly (of items not yet actively used in data taking), but effects could also be too subtle to be detected with limited statistics in test samples.