Bounded Complexity Justifies Cooperation in the Finitely Repeated Prisoner's Dilemma