Optimal consensus for uncertain high‐order multi‐agent systems by output feedback