I think you mixed together two cases here - when there is a single voice, and when there are multiple voices. (Or possibly I confused them. )
What I am saying is that if it were possible for a single voice to use two different libraries, that most likely any dynamics tweaking you did (as one example) would need to be different for the two different libraries. If you have just one voice… you only have one lane which is shared.
I probably should have explained better what I was thinking, but I feel that using (or miss-using?) multiple voices is a long way from easily creating complex multi-library templates. (My emphasis) and I feel like that was not their intention,
The more I think about it, the ultimate reason for me is that Dorico is not a full DAW. I need each of them to export separately and easily - as player staves do - so that when it gets to the DAW I can do things with them that would otherwise be problematic in Dorico.
If I need a big layered sound, I am probably not going to get it with a single track… Yes I can split an instance into different tracks in the Dorico mixer, add their own inserts etc… But I can’t export them separately - unless I mute and export multiple times, rename files, etc. I can’t do things like pan the sends or have multiple sends…
I’m eagerly waiting the DAW/Dorico integration. Or then again maybe not. The biggest limitation may be that I need my processing power back - Unless I want to start a “Go Fund Me Page” for an HP gen 10 blade, I need the instruments to be “printed” and the samples unloaded before I start loading a Fairchild emulation, several convo reverbs, etc.
Again, all just me.