Can the task "process_flow log cleanup" be made flexible, instead of being started every 5 minutes.
If the cleanup has run and the tables have been cleaned, this does not have to happen again after 5 minutes, this can also be done once a day. In our situation, it is a comparatively expensive query.
As a potential solution, you could consider rebuilding indexes as discussed here: