The tension and deflection rate would show up as increased variation (inconsistency). That doesn't appear to be happening. In fact, right now I don't know if the accuracy of the force gauge isn't the single largest error in the system.

Despite hand winding, hand pressing, off-angle winding, different spools of line and other issues, the standard deviation is still down in the 6-10% range, which is typically about as good as can be expected in these tests.

So, at the moment, I have no real motivation to try to improve the arrangement: It's good enough.

I think the type of filament is important in some of the issues you mention. For example, if the line stretches or doesn't stretch, the amount of tension might be very important.