We're not applying a "fast path" for when a FulltextQuery doesn't match anything: there is quite some work which can be avoided in the preparation of extraction of all results.