The line plots cache a sorted array of points, instead of assuming the
points are sorted. This does cause duplication of the points, but allows
for a much faster binary search to be performed. Perhaps there should be
a cutoff on the size of the array to be cached.