New example: Systematic testing of intersection (and approximation) algorithm(s).
The idea is that approximation should generalize intersection. Hence, the example now shows that the same intersections are obtained using approximation.
To also illustrate differences between the two methods to find intersections, the distance between points at optimum are computed also for the approximation solution, and these turn out to be bigger.
It should be remembered that computing intersections is still more efficient than approximation (when one knows that an intersection exists). It would be possible to improve the computation of approximators so that it switches to intersection-finding once one intersection has been found, but this is not implemented at the moment.