Approximating Perfect Recall when Model Checking Strategic Abilities: Theory and Applications