Imprecise Probability and the Measurement of Keynes's "Weight of Arguments"