If it is in principle possible to experience differently from what a quantum scan of the brain and body would indicate, but behave in accordance with physicalism ... how would you know if what you experienced was different from what you thought you experienced, or if what you thought was different from what you honestly claimed that you thought?

Pardon?

How could you tell if you were experiencing something differently from the way a p-zombie would (or, if you are a p-zombie, if you were experiencing something differently from the way a human would)?

But if you are a sentient cat (roll with it) then you have evidence of the epiphenomenal cats, even though this evidence is inherently subjective (since presumably the illusions are also seemingly sentient, in this case.)

In every meaningful way, the cat fairy is a cat. There is no way for an epiphenomenal sentient cat to differentiate itself from a cat fairy, nor any way for a cat fairy to differentiate itself from whatever portions of 'cats' it controls (without violating the constraints on cat fairy behavior). Of course, there's also the conceivability of epiphenomenal sentient ghosts which cannot have any effect on the world but still observe. (That's one of my death nightmares—remaining fully perceptive and cognitive but unable to act in any way.)