SpeedUp plotting

[sqbl]

Currently, the plot_objective is painstakingly slow (especially so at high dimensionality and with many data-points).
Part of a solution could be to reduce the number of calls to model.predict(X).
Additional work should be placed on investigating the time used for the scatter-plot of the experiments - it scales very expensively wrt time when many points are to be drawn.

[dk-teknologisk-mon]

One likely way of speeding up this part of the code could be to replace the random value sampling used to build the plot data with golden ratio sampling (R1 sampling) like we use in the SumEquals constraint function.