OK. Now even I understood it 
Those dirty table reads appear only to occur on the filecache and the storage tables.
This is what I would do if I would get those messages:
Reindexing all files from scratch:
occ files:scan --all
… and if that does not solve the issue, I would once more give this a try:
occ maintenance:repair --include-expensive
… but I do not have much hope about that one, because all repair jobs are normaly executed after an update…
Much and good luck,
ernolf