should we verify the construction of the clues somewhere?

[redshiftzero]

Starting with #1312 we'll be adding clues into transactions. There is not a 1:1 mapping between clues and outputs however. I'm wondering if we should verify the clue generation somewhere (in ZK) to check that the clues were constructed for the right clue keys. Rephrased, we would verify (perhaps in each output proof) that there exists a valid clue in a list of clues for the diversified clue key of the destination address. Does this make sense to do?