On the Convergence of Policy in Unregularized Policy Mirror Descent