Is Cross-validation the Gold Standard to Estimate Out-of-sample Model Performance?