Block out of square

So I've been working on getting my bell dialed in on a crate motor, 408 stroker out of a 360 block.

When I mount the bell, all bolts, properly torqued, stock pins, etc, it's out by 46 thousandths, pure vertical. The bell is high, relative to the crank, by 46 thousandths. I've dialed it repeatedly, remounted the mag base multiple times, in different places on the crank flange, verified that there is no wobble at all, the dial is properly affixed perpendicular. I get exactly the same each time.

Talked to the crate engine builder, they say it's impossible for the block to be that trashed, blame the bell. Talked to the bell manufacturer, they have the post-production numbers off the CNC measuring table for that serial number, says they measured it and it's true. Went around in circles for a long time.

Finally got an idea. Put my dial indicator on the crank face, and I dial indicated the back face of the block that the bell bolts to. Its out of square with the line of the crank. I zero the dial with the indicator tip at the edge of the flange that the bell bolts to, level with the crank. As I turn the crank and the dial works its way up to 12 o'clock at the top of the flange, I get a perfect, linear increase in error. The top of the block is 40 thousandths shy of being square. (i.e. the face of flange is displaced towards the front end of the engine) As I come back down the block on the other side, perfect linear recovery to zero. There's no wave or irregularity in the flange. There's no indication of hand grinding or any other weirdness on the flange. The best theory I have here is that when the block was originally machined, it was improperly clamped, perhaps with debris under the nose of the block, which resulted in the rear flange being milled out of square.

So, assuming I'm not insane, and the flange is out of square with the line of the crank -- Then as the bell goes on and I torque the bolts, the bell settles into place, rocked forward compared to where it should be, which in turn lifts the rear end of the bell, and displaces the centering ring of the tranny upwards. Since the bell is roughly a hemisphere, if the top of the bell moves forward by 40 thou as the bell is rocked forward, that means the back moves upward by roughly 40ish thou, which puts me in the ballpark of the error I see when I dial indicate it. -- But it's not a pure linear shift of the bell, instead it's an angular misalignment which manifests, when I dial, as a linear error. Yes?

The block was align honed, but it seems impossible that the crank itself could be out of square with the block. The crank is about 1.5x as long as the distance from the crank centerline to the flange, so that means if the error was in the crank relative to the block, There would also be a total of about 60 thou worth of difference from one end of the block to the other. That, in turn would make the piston height at top of stroke vs the heads have incredibly large variances from one end of the block to the other, and I also have a feeling that with so much angular error, everything would bind. The engine has been on the dyno and it makes 530 horses and 530 ft-lb. So it seems counterintutive to believe that -- if the problem was the alignment of the crank within the block, that it would perform so well. If there was that much difference, then since each cylinder would have very different compression, different amounts of squish, it should in turn lead to weird combustion differences from cylinder to cylinder that I would think would be very obvious and apparent. That leads me to the belief that the crank, journals, cylinders, head decks, etc, are all good relative to each other, and that the rear flange of the engine is the only likely problem. Measuring piston to head deck seems the best way to verify this, but I'd rather not have to take the heads completely off to verify this, and I can't see any easier path. Maybe testing compression for each cylinder and verifying they all match would be a good way to infer that the piston heights were all matched relative to the head decks and therefore in turn, that the crank was in fact true, relative to the block?

1. Does this logic all make sense? Is there something I'm missing here?

2. I'm thinking my best path to good.
I thought of attempting to get a half-moon shaped shim, the size of the mounting face of the bell, and machining that shim such that there's 40 thou worth of taper in it. I talked to a number of machine shops, each of them thought they had a way to do it, by placing a block underneath, shimming, and then milling the part to be machined flat, introducing the needed amount of wedge, but each, after several months, eventually returned to me saying they couldn't do the work, usually stating that they were unable to determine how to get it clamped sufficiently.
From there, my observation is that the bell itself has a scalloped shape compared to the block. Thus, there's only actual contact in the area around the mounting bolts. Between the bolts, the bell doesn't touch the flange at all. So my next thought it to get a set of precision shim washers, in a range of thicknesses, such that as a set, they make the amount of ramp I need to bring the mounting surface back to true. At that point, I believe my dial indication comes out at or near 0, the angular misalign is corrected, and I should be good to proceed.

Any other thoughts or advice from the community?