Add scoring metrics of initial genotypes to `init` methods of algorithms.

[Lookatator]

As mentioned by @hannah-jan, it’s important to include evaluation metrics of the random initialisation of genotypes so we can see how much the first iterations make a difference.