internal: can set bevel to true and get non-garbage result

bevel is always set by thread depth
   acme takes tpi
   square threads are at angle 0
   added generic_threaded_{rod,nut}
   eliminated metric_trapezoidal_*
   cleaned up matrices some in generic_threaded_rod
   threaded_rod can produce spec-true ISO/UTS profile with a triplet input for the diameter.
   Added bevel1 and bevel2 to all modules.  Made default uniformly false for every case instead of
       sometimes true, sometimes false
   Profiles that go over zero are not clipped, and bevels are based on actual profile top, not nominal
   When bevel is given to nuts it bevels the outside of the nut by thread depth
   higbee looks best with quincunx, but it's more expensive.  Select quincunx when higbee is used, min_edge otherwise
   Current code uses difference to remove excess length in the rod.  This gives faster renders at the cost
      of more complex code and green top/bottom surfaces.
   Changed slop to 4 * $slop.  I got good results printing with $slop=0.05 with this setting.
   Don't generate excess threads when starts>1, and don't force threads to be even

vnf.scad

@@ -216,7 +216,7 @@ function vnf_triangulate(vnf) =
 
 // Function: vnf_vertex_array()
 // Usage:
-//   vnf = vnf_vertex_array(points, [caps], [cap1], [cap2], [reverse], [col_wrap], [row_wrap], [vnf]);
+//   vnf = vnf_vertex_array(points, [caps], [cap1], [cap2], [style], [reverse], [col_wrap], [row_wrap], [vnf]);
 // Description:
 //   Creates a VNF structure from a vertex list, by dividing the vertices into columns and rows,
 //   adding faces to tile the surface.  You can optionally have faces added to wrap the last column