From: Predicting the difficulty of pure, strict, epistatic models: metrics for simulated model selection