Add G4VG support for ALICE Geant4 geometry components

[sethrj]

Add G4VG support for ALICE Geant4 geometry components

[sawenzel]

From discussions with the VecGeom team it sounds like we might be able to reduce the displaced solids to regular solids by propagating the displacement into the "placement" of the physical volumes.

Yes, that's a good idea. In Geant4 a boolean has 2 logical volumes as members whereas in VecGeom there are 2 placed volumes. A logical displaced volume should be straightforwardly convertible.

[sawenzel]

Since you have the code already I would say just go ahead -:)

I believe the assumption might be based on some documentation but potentially the gdml converter or some other unit does not seem to respect it.

[sawenzel]

I made a PR with the replica treatment. Feel free to use it or take it as inspiration. I am now locally able to convert my test geometry.

[sawenzel]

@sethrj : After inspection of the conversion code, I believe you should already be doing the right thing. But there might be a wrong assumption that only the second volume could be a displaced volume. In fact, after switching on assertations, I see:

.../app/celer-g4/celer-g4.cc:297: critical: While running input_gun.json: celeritas/ext/g4vg/SolidConverter.cc:696:
celeritas: Geant4 error: celer0001 failed:
    precondition failed: !dynamic_cast<G4DisplacedSolid const*>(solids[0])

So simply extending the treatment to the first volume might be enough. Shall I provide a PR?

[sethrj]

@sawenzel Yes that was an assumption inherited from an earlier version of the vecgeom->Geant4 converter. I actually just found the same thing and am about to submit the PR! Thanks though. If you'd like to add the PV expansion we'd love to include your contribution 😄