Envy-free Policy Teaching to Multiple Agents