Let's add case geometry to this. And no, I do not understand it myself.
I note when looking at Noslers data that the improved 30-06 has a .2 gain in case capacity over the standard. That's looking at 175-180 grain data. The charges of H4831 and R22 are the same, yet the velocity is higher with the IMP.
I don't get that.
I did read one article in which the man, who claimed to be an engineer, said it had to do with the reflective angle of the neck wall and the apex of that reflection. He made a convincing argument for his hypothesis, but I don't know if I buy it.
The first thing going on here is that most load data is wrong. For any reasonable .30-06AI chamber you'd get a lot more than 0.2gr extra case capacity. The actual gain is somewhere between 2 and 6 grains depending on the two reamers being compared and whether you're comparing to pre or post firing .30-06 brass.
The second thing you've discovered is what's called Sebert's factor, discovered by a 19th century French ballistician. What he figured out is that some of the powder travels down the barrel with the bullet. This acts as if the bullet is heavier (the powder has to be accelerated) but doesn't give you any benefit in terms of bullet velocity. It's just an inefficiency. An ideal cartridge would keep all the powder inside the case for the entire combustion, just venting gas out the mouth. Of course, there's no way to make that happen. For highly efficient cartridges with shoulders that trap powder well, as little as 30% of the powder weight has to be added to the bullet weight. For highly inefficient straight wall cartridges, it can be as much as 75%. The .30-06 is on the inefficient side for a bottleneck (small shoulder diameter, shallow shoulder angle). The .30-06 AI is much better (both parameters are improved).
