Add HillClimb selector evaluate methods

[dbalcomb]

This adds a new evaluate method to the HillClimb selector.

The HillClimb selector is a composition of a selector and a mutator that simply returns the best individual within a set of iterations. The implementation relies on the mutated individuals being correctly evaluated so that the best fitness individual can be chosen. This requires users to call evaluate on the mutator passed in. It may also require users to call evaluate on the selected individual but this depends on whether the selected individual has previously been evaluated. Calling evaluate on the HillClimb type itself is almost never what you want as it would result in the selected individual being evaluated which should already happen within the selector.

This change introduces new evaluate_iterations and evaluate_mutations methods on the HillClimb selector that transform the inner operators. The former adapts both the selector and mutator with the given evaluator, requiring the Clone bound, whereas the latter adapts just the inner mutator with the given evaluator.

These methods are an alternative to chaining on the selector and/or mutator directly but otherwise provide no new functionality. The original plan was to override the evaluate method but it is possible that users will want to hill climb with a separate evaluator and then evaluate again. The evaluate_iterations method ensures that both the selected and mutated individuals are evaluated consistently whereas the evaluate_mutations method assumes that the selected individual has already been evaluated.