You don't generally see that addressed elsewhere because it's incredibly difficult to properly measure, at least not without having all the variables and parameters available to you. Jeff Siewert mentioned this in some of his information. Imagine you have an X, Y, Z earth coordinate system. Z is your horizontal (to the right), Y is elevation straight up, and X is looking down range from the base of the bullet in the direction of fire.
Where the moment arm stuff comes into play is with a second coordinate system that is not earth fixed, but instead fixed to the projectile, we'll use i, j, k. Axis "i" runs through the projectile CG (assuming it's has perfect symmetry), from the nose of the projectile to the base. This is the polar moment of inertia. The other two axis, j and k, run perpendicular to the i axis through the CG, and are known as the transverse moments of inertia.
Here's where your point is important - if the CG of the projectile is offset some number from the axis of symmetry, then you now have an arm running through either the j or k transverse moment of inertia, perpendicular to the polar moment of inertia. That arm now wants to cause the projectile to "spin" within its current state of spinning (if that makes sense). This mass asymmetry (you called it mass distribution) is what leads to lateral throw off, and if I recall correctly, is the largest contributor to what Jeff Siewert described as a "cross velocity". With a pile more math and so on, you would be able to visualize the projectile motion (along the i axis)as a sine wave, depicting the frequency at which the projectile is wobbling.
Remember too, the motion of the projectile is "reasonably" constrained within the the bore of the firearm as it travels down to the muzzle. It's still not technically rigidly constrained in the bore, but significantly more-so than when it exits the muzzle. When it exits the muzzle with an offset CG, and it's all of a sudden not constrained to the bore, it can do and I say this broadly, whatever it wants.
