A real head scratcher type issue.

just a guess: some type of induction heating of the metal milk can from the solenoid coil? Did the can jam in the mechanism for a period of time before it was ejected? Some non-ferrous metals can be heated by an induction coil. I have seen aluminum milk cans with a ferrous metal chip wedged in the bottom. Could a can without the magnetic chip be more subject to the induction heating effect? Again, just a theory. At worst, discard the troublesome can! 

I think the steel wedge is just for weight ballast  (?)

You never mentioned when the hot barrel was ejected, that might have localized the issue. 

  I wont rule out fluke induction , but I'm betting the mechs arm is just hitting a coil wire, maybe just during arm travel. Watch the cloth insulation isnt being pushed out of the way too, etc.. Or if it was the 3or 4th barrel only, the coil wire might be shorting on the ramp bottom heating it and the closest barrel too. Just moving the wire haphazardly might place it in the path of the mech, it's a vicious cycler  

Induction? It would be a one in a million fluke and all you could do is shield it with more metal, maybe move wires away from the ramp a bit, etc.

And/Or your operating track might be leaking voltage causing the coil to always be partially active (giving oportunity for heat through time, though mild induction, or coil heat transfer via proximity/parts.{coils do get VERY hot while working. Coil heat would likely show in barrels #5 and after})

Oh the metal wedge again....Maybe there were thoughts to magnetize the platform instead? 

My guess:

The "steel wedges" are either bad magnets, or Korean war era production. No magnetic material available for toy trains.