A binary compare for duplicates during merging is exactly what I thought would be helpful to avoid all of the trouble I'm having now. So yes, I see your point on the merge plugin side of things.
In the second paragraph you said that you think its unnecessary for a merging plugin to allow a user to look and choose between metadata from either of the potential duplicate books. I think this actually would be a useful option, and a merge plugin like what you described with this feature would solve 99% of my problems before they happened.
I agree with you that the root of the problem is in the merging.
Your plugin is extraordinarily useful nonetheless.