First off, thanks to all the maintainers of this project for the quick response/updates that were made to support the V3 search APIs! Your efforts are greatly appreciated!
It seems that the V3 next_page_token is now used correctly in the synchronous iter() API, however it appears that the async APIs try to use the V2 pagination scheme, which results in a never ending search result set. While this functionality can be manually implemented using the new next_page_token() method on the SearchOptionsBuilder, it appears that the intention was to keep that method "internal".