Noticed while investigating HPCC-25465 - if significant time elapses (on other threads) during the calculation of the new timeout waiting for a flow control message, it can end up being very small or effectively zero.