As nominal stress does not have a precise value (hence its name), it doesn't need a precise definition.
I think of it as force divided by nominal cross sectional area.
Where nominal cross sectional area ignores such things as thickness tolerance and stress concentrations, but often includes subtracting things like corrosion allowance.
Nominal stress could also be the average stress across a nominal cross sectional area taken from an FEM result. This average may include the effects of a stress concentration.
The two examples above are both the nominal stress of the same system and yet have slightly different values. That is acceptable, because nominal stress does not need to be a precise value. Within a few percent is fine.
