Gravar-mail: Estimating parsimonious models of longitudinal causal effects using regressions on propensity scores