I think "in principle" is still better - what it means that you can't observe it directly, indirectly, infer its existence, etc. There's not a single bit of universe state that would be different depending on existence of phenomenon under question (if even a single bit of state flipped, you could in principle observe it).

Qm - I assume you mean wave functions and their collapse - are an explanatory mathematical abstraction, but they follow rules grounded in observational data. You don't say "wave function collapsed there, only it's impossible to measure it"; that's meaningless. We say that the wave function collapses when we find some observation changing from one state to another in a way that implies there were superpositions of state involved in the middle.