When reducing terms aggs results, we check if we already have a doc count error for a certain bucket by looking at its error see https://github.com/elastic/elasticsearch/blob/master/server/src/main/java/org/elasticsearch/search/aggregations/bucket/terms/InternalTerms.java#L245): if it's greater than zero we have already calculated it, while if we have zero it means we have not hence we ignore such value and use the doc count of the last returned bucket.

When performing incremental reductions though, 0 may mean that the error was not previously calculated, or that the error was indeed previously calculated and its value was 0. We end up rejecting true values set to 0 this way.