modify CommonMuonHit to support MC matching
Along with Rec!1711 (merged) this tries to make MC matching work for muon hits.
The problem seems to have been that the linker table generated in Boole uses MuonTileIDs of the uncrossed hits/strips, but in the muon decoding these are combined into crossed hits (with new/different MuonTileIDs) and the original uncrossed MuonTileIDs are discarded. This prevents navigation from crossed hits to MCParticles.
This MR adds a "subtile" concept to CommonMuonHit, which preserves the information needed to navigate the truth tables.
As-is this may be a slight throughput pessimisation, as it makes these object bigger even for "data", but it would be surprising if this was significant. I can check that if people are otherwise happy to merge this. I understand from @rvazquez that some more significant changes to the muon decoding are expected in the coming months, so I think we can view these changes as interim ones.
I also addressed some of the concerns raised in Rec#86 (closed). I think I have made things better without actually doing enough to mark that issue as resolved. As I am changing CommonMuonHit I think it's at least important to simultaneously retire the duplicate version, as these will no longer be identical.
This also makes MuonDAQ build both a normal library and a component library, so that downstream packages can link against the definitions in Muon/MuonDAQ/src/*.cpp.