A retail company ingests point-of-sale transactions into a Fabric Eventstream. The analytics team needs a per-store total sales figure computed for every fixed 5-minute interval, with no overlap between intervals. Due to intermittent network issues at some stores, a small percentage of events arrive up to 90 seconds after their event-time window has closed. The team wants these late events to still be included in the correct interval's aggregate rather than dropped. Which combination of windowing configuration should you use in the Eventstream aggregation?