A per-query population used to compute sparse-vector IDF statistics.
The corpus is independent from the retrieval filter and is usually broader.