We can agree we don't want to suffer involuntarily, but since the suffering we wish to avoid is subject-dependent, it's merely a category of subject-dependent things.

Stance-independent definitions are just definitions all the major stances agree with. "Disvalue" can never be a consistent term, it's just that for language to function we can't iron out the details, good enough has to be good enough. When a term becomes crucial philosophically, and the details matter, the differences are revealed. The differences justify the different philosophical positions or the philosophical differences justify the details for the different parties. And commonly, all parties, in an attempt to give their view some authority, try to claim that authority through language. Involuntary suffering, disvalue, whatever terms you decide on, are insufficient as their mere dictionary definitions. I'm talking about interpretation, application, characterisation, emphasis, consequences, hierarchies of value and the list continues on. All you can do is argue for your stances and explain when it is best and why.

I think I worked this problem out. What you can call preferences are actually 3 things together: the perception of the object that is preferred / disliked, the cognitive process of the preference itself, and the functional property of motivation or repulsion that the cognitive process harnesses.

Obviously the way we perceive is stance-dependent. Some people are colorblind, can't taste asparagus etc.
Also, our cognitive capacities and dispositions are different, so the way our own individual brain processes a perception into something that is either liked or disliked, is stance-dependent.

But, the very force of either motivation / approach and repulsion / avoidance, that is stance-independent. If you say that liking / value / approach motive is dependent on the perceiver, then this leads to absurdity. It means you would have no means of deciphering what people mean when they say "I like x", some people could mean things they like, others what they're neutral to, and others what they avoid. There has to be a consistent definition of approach/avoid, value/disvalue, liking/disliking, even if everyone's dispositions and perceptions are different. People obviously can tell what "liking" means, despite the fact they've learned this definition from many observations of different people liking or hating different things.

Another thing this illustrates is that stance-dependency and definitions are different. I can define perceptions, I can define cognitive dispositions / preferences, and I can define value/disvalue, but only one of those things is stance-independent.

If I couldn't define something, I suppose this would be neither stance-dependent or stance-independent, because there'd be no definition for that stance to apply to.

Seems fine. When people like things, it's always positive. When they dislike things, it's always negative. I'm sure we can muster counterexamples (Stockholm syndrome?) but that's besides the point.

The problem I see is that the OP concerns the actions of one individual and the idiosyncratic preferences of another. Let's take a mundane example. Anna and her sister Barbara have asked you to hang a picture on the wall. Anna likes the picture above the fireplace but hates it above the sofa. Barbara loves it above the sofa but hates it about the fireplace. They only agree on hating it everywhere else. So here the forces of liking and disliking are balanced unless you choose not to act, in which case you maximise disliking*.

Morality concerns how we should behave (practical reason). Moral realism states that some claims about how we should behave are true. If in the above example, one or both sisters must suffer and one or neither of them must be pleased as an outcome of your behaviour, how do we go about finding the pertinent moral proposition that is true and, if we can't, if there's literally no difference between two contradictory behaviours, how do we justify our belief in a true proposition?

*The solution obviously is to shoot Barbara in the head so she can't dislike anything and hang the picture above the fireplace.