Fix #388, #383: concave CD, XYZ inside CD, conestack, and box geometry errors

[rtownson]

1. Fix a special case of geometry error where geometries that are convex individually result in a concave cd geometries when combined (issue CD geometry does not properly account for compound concave geometries #388). These geometry errors could be created using a radius-mismatched cylinder and sphere in a cd geometry. Also fix where a small step is taken in cylinder or sphere howfar routines to step by boundaryTolerance/2 instead of boundaryTolerance. This resulted in checks against the boundaryTolerance to fail in the cd geometry.

2. Fix a geometry error where particles could get stuck on an edge of an XYZ geometry used as the base geometry for a CD inscribed in an envelope (issue Trapped particles in xyz geometry used as cd base, inside an envelope #383).

3. Fix an infinite loop that could occur in long simulations with a conestack with more than one layer. Previously, particles could get stuck on the surface between layers of the conestack. Example that would crash when simulating 1e7 particles:

:start geometry definition:
    :start geometry:
        library = egs_cones
        type = EGS_ConeStack
        name = inner
        axis = 0 0 -2.5545 0 0 1
        :start layer:
            thickness = .321
            top radii = 1.9
            bottom radii = 1.9
            media = air
        :stop layer:
        :start layer:
            thickness = 2.0585
            top radii = 1.9
            bottom radii = 1.9
            media = air
        :stop layer:
    :stop geometry:
    simulation geometry = inner
:stop geometry definition:
:start source definition:
    :start source:
        name    = isotropic_point_source
        library = egs_isotropic_source
        charge  = -1
        :start shape:
           type = sphere
            midpoint = 0 0 -2.5
            radius = 0.01
       :stop shape:
        :start spectrum:
           type = monoenergetic
           energy = 1.1
       :stop spectrum:
    :stop source:
    simulation source = isotropic_point_source
:stop source definition:

4. Fix a geometry error where a particle could get stuck in the corner of an egs_box. The error could be produced using:

:start geometry definition: 
     :start geometry: 
         library = egs_box 
         name    = the_world 
         box size = 50 50 50
         :start media input: 
             media = air 
         :stop media input: 
     :stop geometry: 
     :start geometry: 
      library = egs_box 
       box size =   2.8 2.8 2.8
          :start transformation:
            translation =    0, 9, 9
          :stop transformation: 
          name = my_box 
          :start media input: 
              media = water 
          :stop media input: 
      :stop geometry: 
      :start geometry: 
         library = egs_genvelope 
         name = the_scenario 
         base geometry = the_world 
         inscribed geometries =  my_box
      :stop geometry: 
simulation geometry =  the_scenario 
:stop geometry definition:

[mchamberland]

Are you sure this is the right solution? Something strange goes on in egs_view with this:

This is a simple conestack with mismatched radius. The larger one is completely missed when particles leave the smaller cylinder towards the larger one.

[mchamberland]

This is the geometry shown in the screenshot:

:start geometry definition:
:start geometry:
name = conestack
library = egs_cones
type = EGS_ConeStack
axis = 0 0 0 0 0 1

    :start layer:
        thickness = 1
        top radii = 1
        bottom radii = 1
        media = water
    :stop layer:
    
    :start layer:
        thickness = 1
        top radii = 2
        bottom radii = 2
        media = lead
    :stop layer:
:stop geometry:

simulation geometry = conestack

:stop geometry definition:

[mchamberland]

Oh, I think I goofed! My conestack was the only geometry in the simulation, so when a particle exits the smaller cylinder, it leaves the geometry, which is why egs_view can't see anything behind it, I think. If you inscribe the conestack in a box, then it looks fine. My bad!

[rtownson]

Correct, that's normal behaviour. Thanks for testing!

[ftessier]

Rebased on develop, no conflict.

[ftessier]

Adjusted commit titles and messages, rebased on develop.