Review for NeurIPS paper: Cross-validation Confidence Intervals for Test Error