Supposedly, visual acuity is a tad better than .0035". Evaluating reflections of a distant line through a tube may very well show distortion that's some multiple smaller than .0035".
If that 26" barrel has more wall thickness at the muzzle, they can touch and have the same convergence as the longer barrels. Maybe so because it might be easier to lop a few inches of the muzzle end of a standard parts bin barrel than special machine shorter tubes. Tubes might not be likely to be shortened from the chamber end.
Maybe the convergence is slightly more for the shorter tubes, but it's too hard to evaluate a tiny shift in POI on the pattern board. Maybe the shorter barrel set has more muzzle flip than the longer ones and more needs to be figured in to end up with a similar POI as the 32" set.
I'd figure, once it's known that wall thickness varies, that could well be enough machining to distort a tube. One of the by products of manufacturing (finishing) that might alter an intentional bend or straightness. The eccentric choke example maybe one of altering flow through the opening than creating a curve in the tube. It could be distance sensitive.
Still hard for me to imaging that an intentional bend was created, visualized, aligned and preserved through the manufacturing process.
