From my investigation this morning with Britt and Kyle, it sounds like they've made a couple suggestions that will get you in the right direction.
We've run into these sorts of problems periodically. It's basically a stacking of tolerances, or even an unusual combination of errors. Here is what I THINK might be leading to your problem:
* Ride height - We use a taller spindle than the OEM unit to increase the negative camber gain upon suspension compression. Conversely, it will also create positive camber gain on extension. This means that the higher the ride height that you select for your car, the more positive camber you will see [and need to compensate for by shimming the cross shaft inboard]. We run into this sometimes with air suspension because the customer can easily manipulate the ride height. Usually not a problem, but IF it is combined with a couple of other possible issues...
* Cross shaft orientation - I think you've now addressed this, but our cross shaft has about .125" of centerline offset built into it. This allows a total of 1/4" of camber adjustment just by rotating the shaft 180 degrees. Easy to miss.
* Caster - I understand you are concerned about camber right now, but realize that the upper control arm mounts are not parallel to the centerline of the car. They are angled in at the rear. By using the caster slugs to move the upper arms back to get more positive caster you will also coincidentally add a small amount of negative camber. Not much, but every little bit helps.
* In theory, one should be able to make a set of control arms that would require the bare minimum of shims to set camber. Unfortunately, fabricated tolerances in the 1960's were expressed in much larger terms than they are today. It was not unusual to see the upper control arm mounts being off by 1/4". That's why even the GM factory used shims for final alignment. Today's standards are so tight that many cars don't even have any provision for setting frontend alignment. Good for them, not so great for those of us who want to try an alternative alignment setting
* In addition to the wider tolerances of 45 years ago, we've seen a great deal of "wear" on these GM frontends, usually expressed by the front crossmember being damaged or just sagging over time, to the point where you get too much negative camber. I understand that is not your particular problem, but it helps explain some of the variables we have to accommodate when designing control arms. While a positive camber issue problem can likely be resolved with shims, too much negative camber caused by a damaged or worn crossmember cannot.
Hopefully you've got your particular problem resolved by now, this is just some more info for those reviewing the thread.