Sorry. I meant the linear workflow part.
Do you mean using ACES to go from working or camera space to display space ie. Output - rec.709 or Output - sRGB?
If that’s the case the reason being is you are using the ACES RRT designed to transform the data to the display in the best way it can matching what we as humans would expect to see which is vastly different from just applying a power function to linear data.
In your case this wouldn’t matter because you said you will deliver the content in scene-referred log to the colorist. What you’d see in the viewer with your reference transform layer would only apply if the colorist is also using ACES to get to the display or if you would deliver a final image in display space.
His video is focused on CG renders going to sRGB but the same principles in setup would apply to rec.709 too. It will give you the same result with less setup work and chance of error along the way.
But if you prefer the non display color managed ACEScg workspace approach with Preserve RGB for each element that’s also fine 