Reward Machines for Deep RL in Noisy and Uncertain Environments