Overhauled the implementation of intersect_line_segment_with_prism. It now does the full exact calculation with no assumptions about proximity, as those assumptions were causing subtle errors. I understand this may be overkill, but getting it working with the "simplifying" assumption that the point is always close to the prism seems in fact to be more complicated than doing the full calculation. If computational cost becomes an issue we can revisit to investigate accelerations..
Overhauled the implementation of
intersect_line_segment_with_prism.
It now does the full exact calculation with no assumptions about proximity, as those assumptions were causing subtle errors. I understand this may be overkill, but getting it working with the "simplifying" assumption that the point is always close to the prism seems in fact to be more complicated than doing the full calculation. If computational cost becomes an issue we can revisit to investigate accelerations..