Quantifying absolute treatment effect heterogeneity for time-to-event outcomes across different risk strata: divergence of conclusions with risk difference and restricted mean survival difference