PE, your reasoning makes total sense to me (though I had to read it a couple of times). I am merely reporting what Haist appears to be saying at the beginning of that chapter. His discussion about grain in neg vs. pos makes no reference to printing and my impression is that he is comparing positives and negatives directly read, without printing. Not exactly a practical comparison, but it addresses the OP's question.
If you compare negs with pos, you do see a grain difference and sharpness difference as stated and Grant and I agreed to that. But if you compare an original pos and pos from neg, then the system diverge as I state. Grant and I debated that during the writing and editing of his book(s). This is a very sticky subject that impinges on product sales and quality that is hard to get across to people due to their preferences and also due to very subjective viewing of images.
I used to bring the galley proofs of Haist home here right where I'm sitting and edit what we argued over during the daytime hours in Grants office.
There is no easy answer, or no easy explanation. This is a very complex engineering study that relates to the film and to the process. It is more sticky with B&W where the process is a variable.
I am relieved that you clarified this (sort of). It has been smouldering in the back of my mind all day. I remembered Grant's chapter, but don't have it at hand, so I was reluctant to comment and perhaps misquote. It appears this is a bit more complicated than one might think at first reading. B&W reversal seems to be one of those third rails of photography. Although I have done a fair bit of it, I am no expert, and have many regrettables and forgettables to prove it.
I doubt there are that many people who do a lot of it non-commercially. Perhaps I am wrong. On the other hand, one can get a respectable product without too much effort. Now, back to the collodion lantern slides.