Remarks on Carleman estimates and exact controllability of the Lamé system