On the determination of the nature of defect clusters produced by displacement cascades. Part I. A critique of the two-and-a-half-dimensional technique as applied to small clusters in silver and copper