Towards a critical evaluation of an empirical and volume-based solvation function for ligand docking