Indirect covariate balance and residual confounding: An applied comparison of propensity score matching and cardinality matching