Let the ZEM calorimeters be built without the far beam line

[sawenzel]

This adds two ZDCSimParam switches so a simulation can keep the ZEM geometry without paying for transport to the ZN/ZP calorimeters at +-113 m.

• The ZDC geometry is the +-113 m beam line, its magnets, ZN, ZP and ZEM; the transport cost is the beam line, while ZEM sits at z = 7.6 m.
• buildBeamLine=false skips the beam line, the magnets and ZN/ZP; buildZEM controls the ZEM calorimeters.
• The ZN and ZP construction moves into createZNZP() and the ZEM construction into createZEM(), so createDetectors() shows what depends on the beam line.
• The PMC and PMQ medium ids move ahead of that branch: ProcessHits compares every step against them for ZEM too, so leaving them in the ZN block gave ZEM hits with no energy loss and no light yield.
• The ZN/ZP sensitive volume lookups are guarded so they no longer abort when the volumes are absent.
• 20 pythia8pp events with TGeant3: 276.9 s with the full ZDC, 141.6 +- 1.1 s with --skipModules ZDC, 136.9 +- 1.9 s with buildBeamLine=false.
• 50 events of 20 muons at 200 GeV into 4.6 < eta < 5.3 give 234 ZEM hits with 2167 photoelectrons at buildBeamLine=false, against 230 hits and 2097 with the full ZDC.

https://its.cern.ch/jira/browse/O2-7158