I would argue that, given the advice is to *not* use anything earlier than 3.2 for btrfs due a bug which can result in a corrupt filesystem on powerloss, describing Oracle characterising 3.0 as production ready for btrfs as optimistic is being rather charitable.
Especially as there is likely a disk format change necessary to fix the painfully low hard link limit in a directory which can break backuppc, gnus and (apparently) git.