Number of users and playtimes for D1, D2 etc of each cohort show wild differences depending on what date range is set, even though the cohorts in question shouldn't be affected by extending the date to include less or more days BEFORE the cohort.