Review for NeurIPS paper: GAIT-prop: A biologically plausible learning rule derived from backpropagation of error