At the risk of sounding like I am suggesting fake diamond stacks again, didn't the old Radley Hunter stacks have a spiral cone which caused the embers to be dashed against the inside of the balloon by centrifugal force? Perhaps a combination of that with water being sprayed against the inside walls of the stack would come close to working. The water would probably evaporate immediately upon hitting the walls when the engine is really working, but even if the moisture content of the exhaust stream is high the final equation might be dry enough for the added water to create some evaporative cooling and bring at least the marginal embers down below kindling point. All theoretical and just a thought. In 100 years of steam, they have probably tried everything and the better mouse (er,spark) trap still eluded them. Seems like if there was a solution, they would have already found it, but who knows?