Imitation Learning with Limited Actions via Diffusion Planners and Deep Koopman Controllers