Remarks about global analytic hypoellipticity