Rule 1 just isn’t true, it’s consistently accurate not accurate in general. You can go play a game on a console without any aim assist, and while it’s painful to get used to, you can still be fairly accurate so long as the game runs ok. But, yes, aim assist + a good feeling handling system for aiming are by and large needed or massively preferred.
For the two observations, it’s good you’re looking at consistency, but your own accuracy data there is a problem. For the larger comparisons that were done with ranked it was roughly 10-14%, but across the board closest to 10%, but for your observations you’re citing 15+ on average. We know there’s never been any patches making aim assist stronger, and it’s certainly not working consistently, so the only assumption to make is that the SBMM is throwing you against more accurate (better) players than not, and you can’t account for large enough a sample size with just your own post-match info gathering.
The above issues aside, it still doesn’t work. As, yes, the “X is a better player because they’re ranked higher,” (rank really has nothing to do with it, using the example since you used it and it’s easy to understand) or rather the player with more experience, is an aspect to consider. The only way it wouldn’t be would be for everyone to be using the exact same weapon in the exact same situation. A BR start is close to that (still flawed as that gun is the seemingly only weapon with strong magnetism, so if you’re on controller and your reticule is red you won’t ever miss while someone on M&K wouldn’t have the red reticule and might not take a shot that would guaranteed hit,) but it can’t deal with what range people fight at and the like, the only way to properly gauge that accuracy would be a shooting range with limited rounds, but that has its own issues. As does taking things from BTB with how many more weapons are in that, which just ends with an impossible goal of getting accurate data.
For Rule 2 and the subsequent points, I don’t see how this is relevant. A soft press of a key isn’t something that’s even a factor at play, as Infinite has no momentum based movement like previous entries, so you can rapid press keys to do large the same thing, only difference being finer directional control of a thumbstick for movement, but the movement speeds and the quick dodging is the same. A “quick wiggle” can’t actually dodge bullets from a BR unless you were well outside its optimal engage distance, at which point there’s no argument to be made. Both inputs have the same magnetism (can go see that on a YT vid, I think it was from Fallout Plays, I can find it if need be,) so you wouldn’t miss shots in a very slight wiggle movement unless you were beyond the safe range for which being close would allow the aim assist of the magnetism, which is well outside the range at which reticule slow or stickiness would work (if it ever even decides to activate in Infinite.)
The perfect kills you mentioned, I would say there is something there with aim assist, as it isn’t off 100% of the time, but it’s so weak when it’s on that it can’t be even close to half the reason. I still maintain it’s the red reticule, as you see red you know you’re hitting all shots with your BR unless the target ducks the bullets while the first one is firing, or they get behind cover (could still get shot, netcode and desync are a killer.) If you don’t see the red reticule on a mouse, like I mentioned above, you might not feel you can make the shot that would hit regardless because of magnetism.