The only reason the lists are grouped by cost is that it is generally the most convenient way of grouping, since in order to make interesting choices happen, there tends to be a correlation between cost and strength (though this is a tendency and not at all a hard-and-fast rule). You could really break the list down however you want, even alphabetically.
Ideally, we would rate all the cards on a single list, but 200 kingdom cards is unwieldy to work with.
Rating all the cards on a single list has it's own issues, although it's not ones we entirely avoid already. It's downright weird to try and say things like, "Which is better, Fairgrounds or Squire"? What does it even mean to say that Fairgrounds is better/worse than Squire? They're just so totally different cards. Of course we still have a similar issue with things like, Duke against Vault, but at least similar cost limits the weirdness in most cases (and potion costs, there's few enough of that we can still do meaningful comparisons).
Although, if we DID want to do an all cards list, I reckon it'd be viable, but you'd probably want to put things through maybe two-three rounds. Two of preliminary rough sorts, then from that create maybe ~10 lists of about 25-30 cards with overlap, and use those to make a master list. Not saying it's a good idea, but... it's possible to limit the unwieldiness.
As for the argument at hand: I'm kind of indifferent. I've seen good arguments on both sides. When it comes to having $6 in my buy phase and 1 buy, Masterpiece is there as an option, giving me 1 psuedocopper and up to three silvers. $5? It's there again. I'm comparing it to the other $5's and considering it's worth. But then again you can consider it to just be another quirk of the $3 card. Like, you could decide to put Ambassador in with the $5's to see how it does. The fact it's cheaper and thus more available, but worse with TfB is just an oddity it has in the comparisons.
Ultimately I think it comes down to what WW said. Grouping by cost is an arbitrary convenience we chose. Now that there are things to challenge our (mostly) fixed notion of cost, we might want to consider other relevant groupings, but I doubt we'll find anything quite so encompassing as cost, even with the slight issue of overpay effects.