Efficiency of block designs for simultaneous comparison of two test treatments with a control