As far as I can tell, claiming that you're producing "precision machined" dice is absolutely useless if you're not conducting and publishing statistical tests on those dice. That ends up with a company being the Monster Cable of the dice world--producing expensive products with attached unsubstantiated performance claims. But there's certainly a thriving market for Monster Cables.
Are Chessex dice terrible? Probably. But Chessex doesn't go around claiming that their dice are perfect.
The Game Science argument about the rock tumbler has a point in so far as all it's doing is introducing more source of asymmetry and non-uniformity in the dice. But saying "Our dice are less terrible than theirs" isn't the same as proving that their dice are close to ideal.
Anyone want to sell me some dice that they can provide an accompanying certificate of passing statistical testing?