Yes, I have no problem with refresh tokens expiring, but I wish the RFC did explicitly include this functionality along with a "refresh_expires_in" field, as you wrote.