Potential crash with Flying Edges
There appears to be an issue with the Flying Edges algorithm that can cause a crash on some platforms and some odd results in others. To replicate, load in this data:
Then, run the vtkm::filter::contour::Contour filter with an isovalue of 300. (Actually, lots of middle range isovalues cause the issue.) On some builds, you will get an assertion failed in the MapPointField worklet (in vtkm/filter/contour/worklet/contour/FieldPropagation.h). In others, you will get a bizarre corner go way out of the space.
I think the problem in both cases is that the interpolated edge ids that are supposed to be computed in flying edges pass 4 are not being written in some condition, and some uninitialized garbage portion of that array is later used when mapping fields.