To explain the 18 hour Levi test the argument immediately degenerates into absurdity.
That may be true, but I'm not trying to explain anything, just trying to find upper bounds. What I'd really like to see is a comparison to the highest known chemical potential per mass/volume, assuming (yes, very
optimistically) no engineering constraints. I'm having trouble moving beyond handwaving on that point, at least in the small amount of time I put into it. I was guessing something like high explosives might be a good starting point, but there I'm not sure whether that's more a question of energy release over a short time than actual energy density, and chemical potential per mass or volume doesn't seem to be something that jumps out of data on a given material. Oh well.
Imagining a black box weighing 90kg filled with unknown materials that generates heat is nothing more than a bad question on a first year chemistry homework assignment, not an explanation of what Rossi is doing. As someone who has graded such quiz questions thousands of times I can assure you that it is not interesting in the least.
Quite true, except of course in the case that one actually has a box of unknown materials that is producing heat, which oh look we do. With all due respect to your no doubt impressive quiz-grading prowess, I find your reasoning there poor. The question of how much chemical potential such a box might contain seems quite interesting to me in these circumstances, as it could set some upper limits on how long it might produce a given level of heat by any chemical means -- but again, YMMV. Feel free to not be interested, as I'm relatively uninterested in the "fraud vs. delusion" debate.
n*kBolt*Te = B**2/(2*mu0) and B^.25 loss scaling? Or not so much? Hopefully we'll know soon...