As the title says. Introducing this test unearthed a bug in the sorting of individuals in the case of multiple objectives (used with hurdles) which this PR also fixes: individuals should be sorted according to their fitness values, starting from the last objective, whereas previously they were sorted starting from the first objective. Behind this underlies the assumption that later objectives (in terms of hurdles) are more informative about the quality of an individual.
As the title says. Introducing this test unearthed a bug in the sorting of individuals in the case of multiple objectives (used with hurdles) which this PR also fixes: individuals should be sorted according to their fitness values, starting from the last objective, whereas previously they were sorted starting from the first objective. Behind this underlies the assumption that later objectives (in terms of hurdles) are more informative about the quality of an individual.
closes #342