It still disables parallelism. No changes there. If you have anneasily reproducible benchmark on a public corpus, I could try to provide an analysis. (Note: if the corpus is small enough, it might not be possible to witness a perf difference. Or it may even be possible that using parallelism is slower due to the overhead of spinning up threads.)
Perhaps the difference is then explained by my choice of search term. The term I tried after upgrading must happen to appear early in the sorted corpus.
I just now tried it with a very rare term, and it does indeed take longer overall to complete the search.
