Ribbon normal vectors do not have length 1

[Tom Goddard]

Ribbon normal vectors do not have length 1. Observed this with pdb 5isz when exporting gltf, an online gltf validator reported normals with length squared 0.97, 0.98..., printing out from ChimeraX confirmed this. Ribbon tethers have normals with length squared 0.04, 0.17, 0.46.... (showing residues within 5A of ligand).

All the normal vectors should have unit length.

[Tom Goddard]

Very likely the tether normals are messed up because of the graphics instancing code applying a skewing matrix (to resize the two cone axes by different amounts) and not computing the normals correctly -- that probably has nothing to do with ribbon code. Bu the normals on the ribbon itself are something computed by ribbon code.

[Conrad Huang]

Fixed in 7f1ac292e by normalizing normals when creating ribbon drawing.