Vision-based Navigation with Language-based Assistance via Imitation Learning with Indirect Intervention