Existence of an optimal control in infinite-horizon problems with unbounded set of control constraints