Non-fragile containment control of nonlinear multi-agent systems via a disturbance observer-based approach