We wait for count to finish before continuing which makes us vulnerable to the way we count not hitting an index, new variations of mongodb that pretend to be mongodb but really aren't (usually things like time series, serverless, data federation..) or just regressions in the query planner.

Perhaps we need an intermediate state where the page of docs arrived but the count is still loading. ie. turn off the count in the tab and the 1 to x of y and the pagination buttons, etc, and at least display some documents.