Planning Over Multi-Agent Epistemic States: A Classical Planning Approach