> I guess one uses busses for various sub-mixes and selects
That is how I currently do it yes.
> If not, please explain how to go about preparing exports
However if/when a good surround ambisonics panner get implemented I
would be doing mixdowns to listen to in multiple different spaces with
different loudspeaker setups in each space. Thus it would depend
primarily on the routing in the panner and the loudspeaker locations.