This fixes #146 by adopting a simpler jittering strategy in the collinear case
Sharing for discussion, because this "fix" is really feeble: if we increase the radius a little bit more (to 5e-8), another test breaks. I'd like to find a more robust approach.
Maybe at some point we'll need to create a fake voronoi diagram from scratch, rather than replay with jittered points, but it seems that this is working well enough for now.
This fixes #146 by adopting a simpler jittering strategy in the collinear case
Sharing for discussion, because this "fix" is really feeble: if we increase the radius a little bit more (to 5e-8), another test breaks. I'd like to find a more robust approach.Maybe at some point we'll need to create a fake voronoi diagram from scratch, rather than replay with jittered points, but it seems that this is working well enough for now.
closes #146