A note on the “value” of bounds on EVPI in stochastic programming