On rules with existential variables: Walking the decidability line