I have put together a graphic to capture where my thoughts are currently going:
Here the small molecules are molecular hydrogen, the red particles are protons and the blue area at the left is the surface of a nickel metal grain. A spark discharge is underway between this grain and another grain nearby (not pictured). Prior to the discharge the two grains were electrically isolated. The discharge is drawing the protons into a recess in the metal grain, where a great deal of pressure is building up. If this process happens sufficiently quickly, e.g., before there is a lattice dislocation, perhaps the pressure could get quite high.
Here is where I think an insight of Ron Maimon's might be applicable . If the pressure is high enough to cause a d+p reaction, or perhaps even a nickel proton capture reaction, normally one would expect a gamma. But the emission of a gamma is a very slow process. Because the environment is electron-rich, and because the electromagnetic interaction proceeds quickly, under suitable circumstances it will be competitively favored over the emission of a gamma, which will take a long time. If so, my take on and adaptation of Ron's insight is that perhaps the [pd]* or [pNi]* intermediate state will couple with the surrounding electronic structure and possibly positively charged lattice ion cores, and the energy that would normally go into the emission of a gamma will instead be divided among a large number of recipients and result in a bath of low-energy photons.
A counterargument can be made that radioisotopes found within metals commonly emit gammas, so there is no reason that coupling with the electronic structure should happen in this particular case instead. I do not have a strong response to this complaint and only observe that the emission of a gamma from a metastable radioisotope seems to be a sufficiently different situation from the decay of a short-lived fusion intermediate state to suggest that something else might happen in the latter case.
 He's looking at a very different system that involves palladium and deuterium, and he makes no claims in connection with the current discussion. Do not be distracted by Ron's reputation of 1—this goes back to events unrelated to the quality of his posts, and previously he was one of the highest ranked contributors at physics.stackexchange.com.