I never said the numbers or fractions of any errors is "huge" - please don't put words in my mouth.

I can't say any part of the fork WILL or WILL not be off by 2 mm. I'm just hypotheticaly discussing how the testing protocols could introduce a bias in the product, in favor of releasing products that could have longitudinal variation but do not have lateral variation.

Again, I'm not an expert in any kind of molding. I have seen some surprising things go wrong in molding of some some smaller parts with high tolerances and complex shapes.