Add utility to estimate nadir point

[kjappelbaum]

Feature description

The computation of hypervolumes is always a bit arbitrary. [1] Shows how the performance comparison depends on the choice of the reference point. To allow for a more consistent comparison, one could always choose the nadir point.

[1] Ishibuchi, H.; Imada, R.; Setoguchi, Y.; Nojima, Y. How to Specify a Reference Point in Hypervolume Calculation for Fair Performance Comparison. Evolutionary Computation 2018, 26 (3), 411–440. https://doi.org/10.1162/evco_a_00226.

Implementation idea

This should maybe be the default for the reference points of our get_hypervolume function?
But then we still have ambiguity over the following questions:

• about which points do we take the minimum
• how much worse than the nadir do we make our reference point?

[github-actions]

Stale issue message