Add support for stochastic objectives

[eugenioLR]

Many objectives in real-world scenarios are not deterministic, and might not give the same objective value in different runs for the same input. We need to research techniques to mitigate this stochasticity in the functions.

Stochastic objectives

• Add flag to disable caching for fitness (fitness is recalculated always before selection)
• Repeat objective function and obtain mean/median with option