Optimization and Clear definition for the Tile-Muon coincidence (ATR-21107)

To optimize the memory usage if the text file of LUT is loaded, and for further development with Run2/3 Tile-Muon coincidence, the several updates and code cleanup is needed. I implemented below:

• migrated TileMuon coincidence map from multi-dimensional arrays to std::unordered_map, as same as TGCTriggerDbAlg::TGCTriggerData
• separated class of CW map/LUT for Run-2/Run-3 Tile-Muon coincidence (for instance, the contents are same.)
• Run-2 TIle-Muon coincidence map is read from CONDDB only, because the DB content is valid and the text file is not available on the repository.
• The propagation of LUT object is improved. Now the each CW map/LUT is set in TGCSectorLogic class, instead of TGCSector by pointer propagation.
• The switching of Tile-Muon coincidence has been fixed and TILEMU=true is now the default configuration. (the configuration is available for either Run-2 or Run-3). The user could add an option TILEMU=false, when the Tile-Muon coincidence would be forcibly disabled.
• other improvements for C++ coding

Run-2 Tile-Muon coincidence has been finalized (although is still possible to improve the coding...). For the time being, the same logic is applied for Run-3 simulation.

test_trig_mc_v1Dev_L1SimOnly_phase1_build.py and test_trig_mc_v1Dev_L1SimOnly_build.py runs successfully.