Update: While I still need the Atmos binaural downmix capability, I am getting pretty close with a work-around:
On the mix bus 7.1.4, after the renderer plugin, I insert DearVR Monitor set for 7.1.4 input, and the output is set in its “control room” choices to “analytic position”. That gets somewhat close but is a tad less present. So I added Ozone EQ Match and matched a great reference I got from a friend. He gave me an AMD broadcast wave file that I generated the binaural downmix from, and he also included the Atmos binaural re-render. With that to work with I got a pretty close match. You’d hear a little more realism in the Atmos one, they are crazy good, but… this seems like it can actually work. I believe only engineers would notice the difference for the most part, and mainly side by side comparison. The panning is feeling the same as though if I deliver an ADM Broadcast File that I mix with referencing binaural in the way I describe above, and it later gets downmixed on an Atmos Renderer down the line, I might just have it. I still want to get this 100% right, though. Waiting on feedback from my friend who is a stellar producer/mixer/mastering engineer. He here’s everything, I’m sure he’ll point out some things, too.