Stackelberg Meta-Learning Based Control for Guided Cooperative LQG Systems