Some remarks on the use of the effective medium approximation for resistivity calculations in liquid noble metals