Why is turbidity more useful in measuring contamination of liquids by large numbers rather than small numbers, of bacteria?