Radar,
I am just going to repeat you to make sure that I understand where you are coming from.
You (your company) has a contract with a sub, and the sub has the leeway to do 'anything' as long as it meets both the performance criteria _and_ code. The sub has come to you with a proposal which, as far as you know, meets both the performance criteria and code, but does so in a fashion that you are leery of. I presume, then, that you want to achieve one of the following:
1) Find a code reference that makes this proposed design untenable, so that the sub simply can't do it the way that you don't like.
2) Find a method that is the equivalent of what the sub is proposing, but addresses your safety concerns.
3) Find a method that is almost the equivalent of what the sub is proposing, with a slight change in performance criteria that will be easy to negotiate.
4) Decide that you can't force this issue and give up, but at least you know that you tried.
5) Decide that you can't force this issue and re-negotiate a different set of performance criteria with the sub so that they have no reason to use the design that you don't like.
IMHO, your best bet is option 3. I do not see a _significant_ hazard in this installation, as long as the wires from the two different systems are very well segregated. If the conductors from system A simply pass through the JBs for system B, and vice-versa, and if the colors for system A and system B were easily differentiable, then the chance of error will be very slight. Since the sub has no reason to cut the wires from one system in the JB of the other system, this is not going to be a significant performance requirement. However I think that you could only make a code case for identifying separate neutrals.
Identifying _all_ of the wires is IMHO a small performance change. I suggested using romex as a _cheap_ way of identifying all of the wires; but perhaps one could use E57's technique of using custom colored wires, or perhaps pre-twisted wire sets. This sounds like a large enough job that the cost of custom colored wire would be a small premium; in fact you might consider doing the _entire_ job with different colors from each supply. Since this is a change in the performance criteria, I'd presume that you would have to negotiate it, and would have eat the materials cost increase. But I'd think that the cost increase is probably slight.
These latter approaches would result in essentially the same labor as currently proposed. Wire manufacturers can easily make different colors or standard colors with custom striping, and can also easily duplex/triplex/quadruplex your conductors. If you absorb the change in materials cost, this might be very easy to negotiate with the sub.
-Jon
