Last unofficial information I have is that this issue is hopefully fixed and the fix will be in v6.9-rc1 which is expected soon™, no idea if a patch/update for v6.8 will be available, but very much doubt that as LT doesn't usually patch older releases after a newer one is in active development, maybe if it was a critical security issue.
Also, and completely by accident I might have found a "fix" for current users, at least it's working for be and it's repeatable, this is iotop after 5 minutes with just the plex docker installed (which appears to be one of the most active):
The "fix" is using the mover to move the docker image to the array (docker service must be stopped first and share set to cache="yes") then back to the cache device (set share to cache="prefer"), start the docker service back up and this is the same docker image running for 5 minutes after the move:
So a 10x decrease in write activity, please try and let me known if it helps.