Short short answer: because those voltages are "nominal", not what's really observed in-circuit. The available output charge (ampere*time -> charge) is much more controllable during manufacture, as it's directly related to the quantity of available reactants within the battery.