After running a half-tank of fuel through this thing just idling, and carefully methodically collecting crazy amounts of data, graphing in excel....i'm stumped!

First issue I solved (tip for anyone else with an aftermarket throttle body!) is to check and if needed, open up the IAC air passageways to and from the seat area (careful to NOT modify the seat itself since that is where all the controlling "valve" action takes place). At first, once the IAC steps reached around 120, no matter how much higher they went, flow was choked and no more idle contribution was actually being given beyond that. After ensuring all passageways were at least the same cross sectional area as the seat, it can now use full range of the IAC steps (although from my testing, I find that after about 200 steps, the airflow gains per step seem to be smaller than what the factory IAC area table shows).

As a preface to the issue that has me baffled, I did a quick check of how sensitive the engine was to IAC steps. Using the DVT tools, with timing locked and ensuring measured Lambda was consistent and cooling fan constant on, for a given TB set screw position I varied the IAC counts and saw (averages from data):
1 step = 625RPM, 7.25g/s airflow
45 steps = 1080RPM, 11.75g/s airflow
All steps in between fell in between as one would expect, so it seemed clear that the IAC effect is significant in this application. However, the issue at hand...

With DVT still keeping everything else consistent/locked, if I let the IAC control idle speed, one minute the engine can idle happy at 15 IAC steps, but if I shut the engine off and restart it, to idle at the same RPM and airflow now requires around 135-150 steps no matter how long after startup I wait?!? Keep in mind, I am keeping timing locked, cooling fan load constant, no steering or alternator loading changes, trans fluid at full temp and in park, and am ensuring ECT, IAT, and Lambda are extremely consistent. I surely thought a vacuum leak or sticking throttle had to be happening, but have double and triple checked and confirmed that's not the case. This phenomenon is very repeatable though; if while idling at 135 steps I gently open the throttle to 0.4 to 0.8% just enough that it doesn't kick out of idle mode, the IAC steps do come back down to around 15 and when I release the throttle and it goes back to 0.0% the IAC steps do NOT go back up, they stay down however the engine speed and airflow are still normal. I can physically go push the throttle arm closed to make sure its not sticking, and even blip the throttle multiple times, and it still is just fine at only 15 steps again. That seems impossible?!? Sure enough, if I shut the engine off, and restart it, when it levels out its back to 135+ steps again. This cycle between 15 and 135 steps is absolutely repeatable and has me baffled. While it is supposedly at 135+ steps, if I use DVT to command the steps lower and lower it does pull the rpm down and stalls the engine, so either there is some mystery airflow getting to the engine or its for some reason reporting 135+ steps when its really at something much less?

Anyone ever seen anything like this? Is the IAC motor maybe not finding its zero position correctly after shutdown/restart or something? I thought that the IAC system was designed to automatically re-zero. With that wild of IAC step range there is no way to keep the idle trims reasonable.