I may have to give your work around a shot. I've been using iotop to watch things and found the official PMS Plex docker was causing a lot of writes. I move the the linuxserver docker to see if that helps, but something was still causing loop2 to write a bunch. PMS just made it a lot worse. My plex DB isnt even on my cache disks, its on a separate SSD via unassigned drives.
I have running a XFS array with a BTRFS cache pool and it unencrypted, so the issue isnt isolated to encrypted pools strangely.