You misunderstood my question about frame averaging. I wondered why to use high ISO with frame averaging. For example, I wondered why you would use 10 frames with 1 second per frame (high ISO) instead of 2 frames with 5 seconds per frame (low ISO). In both cases, the motion blur will be the same.
You would not use frame averaging to avoid motion blur, on the contrary.
I can only speak from my own personal experience in cities and urban landscapes, but if I have a high contrast scene exposing to protect highlights and where I know that I’m going to want to pull out deep shadow detail, then the more frames I have in the calculation, the cleaner the shadows (which I know is obvious). This can also be useful if I have to stop the FA in the middle for whatever reason. Also, as often happens in urban areas, when light intensity can change quickly, especially at night with vehicles going by or any variety of lighting changes as the city settles in to the dark, I want to avoid "additive" components of any exposure. If the sensor is open long enough to blow the well, it's blown out. Fewer images with overblown highlights can have the affect of overblown highlights in the averaged image.
I can choose to get those extra frames a couple of different ways, either by extending the length of time I record or, if I'm time limited, increasing ISO (to a point), assuming I want a constant aperture. The caveat is that once the shutter speed exceeds the sensor scan speed, then the number of frames that can be captured in a specific time interval becomes constant which is where you end up with gapping. There's a fairly quick point at which increasing ISO doesn't really help in the "maxing out the number of frames" scenario, for how I shoot.
To keep the math easy, if not always realistic, I can get 120 shots in a two minute frame average at an exposure of f8, ISO 100, 1 second, I can get (because of the above caveat) an upper limit of ~275 at 1/4 second at ISO 400 over those same two minutes and the resulting raw file is super clean with no real penalty except maybe gapping. Whether that matters is scene dependent. I’d need an additional two minutes or so to get those frames at ISO 100. That may seem like a petty amount of time, but sometimes being able to move scenes reasonably quickly is of value to me, especially as light and other factors around me change.
Do I really need those additional frames to pull super clean shadow detail? In most cases, probably not, but having that flexibility is nice. In my early testing a couple of years+ ago, I did find there to be differences in certain scenes as to the number of images per interval as regards averaged noise in shadows. Because I've matured over that time, it may be that I wouldn't find it to be true now. To me it's less about equivalences, but having more than one way to get there. That may not be important to many people but I appreciate it, nonetheless.
To be clear, most of the time, if I do use AFA (which isn't that often), I do shoot at ISO 100 because my shutter speeds are typically reasonable and I hit close to the max limit of frames per time interval anyway. The key for me is that there’s flexibility when I want it. I’m not invested or interested in proving AFA is the best thing since sliced bread, or that anyone should run out and pick up an IQ4150, or that if they don't have one, they're not going to be successful. Clearly, that's not true and some cameras have a variation of it, and honestly, I don't have the back because of AFA. It just showed up in my firmware one day

. AFA is just a tool in the tool belt, as was noted elsewhere. I do appreciate the flexibility.