Scale-free adaptive planning for deterministic dynamics & discounted rewards