Convergence and asymptotic agreement in distributed decision problems