I know my state tracks case data and death data by date of test/death, not by date of report (which generally means that the last ~2 days in the daily report are undercounts, although annoyingly, the running averages don't stop a few days early).

I don't know how well other jurisdictions do at distinguishing between day-of-report versus day-of-event, nor how well aggregations do at tracking the disparities here.

In any case, at current trends, "3100 people died in a single day" will shortly be true if it is not true already.