Add CSG solids

[sethrj]

CHAINED on #1142

This adds basic solids, which are shapes with interiors optionally subtracted, and azimuthal sections optionally removed.

[sethrj]

@elliottbiondo This is the end of the queue!! 🎉

[elliottbiondo]

Since this is the last PR in the queue, could you add a brief design document? Maybe you did in a previous MR after I already approved and didn't see it.

[elliottbiondo]

I would say "interior excluded region" and/or a wedge subtracted.

I also don't understand why it couldn't be neither, though I see that that is how it is coded.

[sethrj]

I'm deliberately trying to prevent two Object types from being the same representation unless they compare equal... maybe I'll relax that eventually, but I think for now I don't think it's a problem to keep the restriction. Or do you think it's better to have a "Solid" that's the same as a "Shape", and later if we need to we can add a "simplify" method that'll let solids decay to shapes? That would remove some logic from the Geant4 translator...

[elliottbiondo]

So if a geometry consists of only simple cuboids for example (with no internal regions or wedges) then solids are never created? In a sense it would be nice to have geometry elements go through the same levels of abstraction, but it might not be worth it if that adds a bunch more code that is essentially a no-op. To weight in further I think need so study the forthcoming design diagram/document, because I am not clear on what exactly happens on each level of abstraction.

[sethrj]

just Shapes, not Solid. The abstraction is that they're both Object interfaces. So in Geant4 a "cone solid" could produce either a ConeShape in its simplest form or a ConeSolid if it's truncated.

[elliottbiondo]

Okay I see. The nomenclature here is difficult especially since Geant4 nomenclature clashes with ORANGE nomenclature (recognizing that may be unavoidable).

[sethrj]

Well, I've tried to delineate it a bit: "shape" is going to be for the simpler objects, "solids" are the weird Geant4 ones. See #1078 (comment)

[sethrj]

@elliottbiondo take a look at doc/main/api/orange.rst... that part describes the user-facing parts and some of the construction logic. I've got most of the basic "proto" construction written which is the final piece of the puzzle... perhaps we could do a detailed review then?