After buckets became always compressed, the query-based reopening pipeline hasn't been able to exclude buckets already with gTimeseriesBucketMaxCount measurements.

We should add this to avoid reopening full buckets. The filter used for uncompressed buckets should still be kept.