I don't think you are right here. It is not just a convention, but a premise. The intension of the formula is to calculate the H or B field at a distinct location ( the center ) in the solenoide.
Of symmetrical reasons you can only find the H and B fields at the center: Amperes law just tells you what the mean values will be, following the circulation path. Now calling the lower left corner of the rectangle ( circulation path ) in the figure in #3, A, then clockwise the other corners B, C, D. If AB and CD are very close to each other, the H fields parallel to AB and CD will be zero, because the H-field will be perpendicular to AB and CD due to symmetry. Making BC and DA longer ( or not centered ), the H-field will no longer be perpendicular to AB and CD. The lengths of AB and CD are assumed infinite, so that the H-field along BC is zero.
Therefore the formula can only find the H-field at the center of the solenoid, and thus DA must be kept very small. That's why n = ΔN / ΔL, ( not N/L ).
I know that the result will be the same, no matter if you use n or N. I'm just speaking of the premises as for the formula, and that the H-field is calculated at the center of the solenoid.
http://h2physics.org/wp-content/uploads/2010/05/solenoid2.jpg