nds_selection: Don't compute hypervolume contribution in a greedy manner

[juliambr]

The computation of the hypervolume contribution (see here) is "greedy". This should not be done greedy, see for example this publication.

Just replace it by the implementation of the emoapackage? emoa::hypervolume_contribution

[juliambr]

Update: also the emoa implementation of the hypervolume contribution can only be used to drop a single point (see documentation). So the issue is still not solved.

I did not found an implementation in R, which efficiently solves the subset problem of selecting a subset of points which have the minimum hypervolume contribution.

[berndbischl]

then this here seems to be the best place for now. but have me review the code at some point

[MLopez-Ibanez]

See: https://mlopez-ibanez.github.io/eaf/reference/hv_contributions.html

It is still greedy and it is not the fastest implementation possible, but it should be faster than emoa's implementation.