A Data-Centric Perspective on Evaluating Machine Learning Models for Tabular Data