A few missing points, which may break the whole plan:

1. For a long long time, it's probably better to increase hedons experienced, then absolute amount of hedonium, and only after there's a whole lot of it do we worry about density.
2. Your description of neural hedonium has no reason to believe it only or primarily experiences POSITIVE hedons.  Your proposal risks creating and maximizing sufferonium.  In fact, this is probably a bigger unsolved problem than identification of substrate.  How do you make something that actually maximizes what you want?
3. it's a bit suspect to go from "play it safe" on the hard problem with artificial neural material, without playing it even safer, and already-solved manufacturing of just making lots of organic brains.