O(N¹/²) convergence in the finite state restricted risk component sequence compound decision problem