On Evaluating Molecular-Docking Methods for Pose Prediction and Enrichment Factors