I think you’ve got the gist of it. There’s no formal concept of an instrument at the pedalboard level, you’re dealing with signal channels and things that operate on those signals. A snapshot (preset) saves the whole pedalboard so you’ll need a snapshot for each unique combination of instrument presets as you’ve identified.
I’ve used my Duo with varying combinations of guitar, mic, and midi keyboard. A challenge is that one can quickly use up available actuators for switching between snapshots. I have a 4-button foot controller with exp. pedal that I use. What’s worked well for me is to have 2 or 3 main snapshots (often for verse, chorus, lead/intro) The other footswitch or two I might use for a specific plugin, switchbox, or override. My Duo is on the back of my (physical) pedalboard so I don’t rely on its footswitches as much, but will often have another plugin mapped and often a mute/kill switch for the mic which is handy.
As the pragmatic cheerleader here, I have to point at out that the current interface for editing and managing snapshots leaves a lot to be desired. The list for managing snapshots is very basic and cumbersome to interact with as you add more snapshots to a pedalboard. If you’ve created several snapshots and decide to change something about your pedalboard like add or replace a plugin, you’ll have to edit all your existing snapshots to make sure they are in the expected state relative to the new change. The snapshots feature is a powerful part of the system but the user experience around editing and managing is a little rough at the moment.