You have said many times that we can just claim convergence, but you fail to mention that this has been answered many times. Yes, there are datapoints that do not fit as they should in the nested hierarchy of life, but the overwhelming statistical evidence indicates that the trees we are finding are real. A clean nested hierarchy can be messed up by convergencies, genes that vary rapidly, patterns deteriorating into randomness in years, and other things. But in spite of this, we consistently find strong statistical significance.
You make a big deal of your vehicle tree with three "taxa": cars, vans and trucks. Do you not realize how trivial a tree with three taxa is? (The stem at the bottom of your tree means nothing, so we deal with this as an unrooted tree with 3 taxa.) It turns out there is only one way to arrange an unrooted tree with 3 taxa. You can mirror image it, for instance, or move the line that says cars over to the other side, but those are just cosmetic changes. The tree is still mathematically the same. So finding evidence for this "nested hierarchy" is like flipping a trick coin with heads on both sides and standing back in amazement when in comes up heads. Your odds with that trick coin are the same as the odds for an evaluation of those 3 taxa yielding your tree, 100%! There is only one unrooted tree you can build with three taxa.
The problem is that each of your taxa is quite varied on many parameters. And when you try to subdivide down to individual "species" of vehicles, you find no way to build a consistent nested hierarchy. Pitabread has done an excellent job of showing how varied the trees of vehicles can be when looking at different sets of criteria. As he pointed out, the trees are coming out with P aprox. equal to 1, which means there is a 100% chance that the cladogram shown is not reliably correct. Contrast that with the multiple studies of living organisms, which regularly yield P<.01 (less than 1% chance that they are not reliably correct), which is quite good for statistical studies.
And no, adding more criteria will not fix your problem. As we have explained, there are multiple ways to divide up cars, and they are not based on a fixed hierarchy. The more independent variables you add, the more hopelessly lost you will be at finding a single correct nested hierarchy for cars.