Hi Huaxin, If #3803 is being closed in favor of a new handler-level PR, I'd prefer to leave my review on #3803 as-is. The existing 'Request Changes' on #3803 has no effect on closing it or on review of a new PR; that PR can be reviewed on its own merits.
As noted on the mailing list, I'm happy to review the new PR once reproducible benchmark results are available. Best, Robert
