Triangles of contact envelopes

[matthiasfabry]

When modeling high-mass-ratio contact binaries, usually a high ntriangles is needed to accurately represent the small secondary component (and the neck region). However, using a single ntriangles over-samples the large primary, which (arguably unnecessarily) causes slow-downs in computes.

I believe this is not unimportant since many contact binaries are found to have extreme mass ratios.

for q = 0.2, the default ntriangles is not enough.