Offset Angle on VPI Classic Tonearm

Actusreus
I suggest ensuring that you have zero azimuth when you are aligning the cantilever with the MINTLP.  If you have azimuth other than zero, either cartridge tilted to the inner or outer groove, you might be twisting the cartridge to compensate for a visual artifact of the azimuth.  I use the MINLP surface, and the azimuth rod, and an accurate ruler to ensure I have equal distance from the rod to the MINTLP glass before I align the cantilever.  Both of my cartridges stick out a little from the front of the head shell, but for all practical purposes the cartridge is parallel to the face and the screws are also in the same place in the slots of the headshell.

That's a great point, Captain_winters. I'll recheck. Thanks!

Hmmm, I thought I posted last night...I spent four hours realigning the Delos yesterday, and it still ended up twisted in the headshell. The azimuth was as balanced visually as it could have been, using the VPI-supplied rod, and a tiny bubble level placed on the headshell. I suppose as long as the cantilever is "parallel" to the reference lines on the null points, it's what really matters.

I'll mention a couple of other things, even though I am sure you are already doing them. I find the 3x much easier to use for viewing the cantilever than the 10x. Too much distortion at the outside of the lens for the 10x for the cantilever. I keep both eyes open. I line up the lines from the reflective mirror first exactly while simultaneously looking at the cantilever. I take the most extreme angle I can, as far forward before the reflective lines separate to get a much view of the cantilever as possible.
As far as lining up to the geometry arc, I find the best way to align the geometry is the 10x on the inside, left channel, of the tonearm. With the front of the 10x loupe round cylinder unscrewed. Just place the 10x loupe on the MINT, on the inside, half inch away. Use the farthest out you can go. The loupe won't roll because your turntable is perfectly level. Lean over and you will see the stylus at the bottom of the loupe, don't even have to touch the loupe. Next point go all the way to the spindle. So the tonearm almost touches the spindle, loupe on the other side of the spindle, still half inch or so away. Lean over and look, from the side of the turntable. Using those 2 farthest points the accuracy increase in getting right on the line is noticeable. So farthest points, almost off the MINT to begin, then almost touching the spindle to end with 10x loupe to check, don't touch it, just lay it half inch away, and you will see the stylus towards the bottom. Since you just aligned, I bet if you check your geometry right now, with these extreme points you will see room for improvement. I have found this method exceptionally accurate.
The cantilever is the cantilever, as long as those reflective lines coincide when you stare at the cantilever to ensure it is parallel, and you are using the 3x not the 10x, then it sounds like you have it parallel, regardless of what the body looks like.

Additional comments.
With such a precise method for aligning to the baerwald arc, you need to ensure you know your optimal SRA first. Which is an entirely different conversation.
The MINTLP is the thickness of a 180 gram record. Standard records minus 50 units on the VTA tower.
The common found digital scale, that goes to the third decimal place is 385 higher than a record.
So for example, my current optimal VTA is 9 full turns on the VTA tower, or 900.
Align with MINTLP at 900. Measure VTF add 385 for a total of 1285. When playing 180 gram records set to 900, for standard records 850. Always check VTF with the digital 3 decimal point scale at 1285.