take your point lyle, but in that instance, a neutral failure would not automatically mean a caravan chassis would become live, you have assumed the caravan has its earth and neutral connected at its incomer, this would not be the case, it usualy would just have a flex of some description feeding a small consumer unit. so the problem would not occur, also the RCD should have tripped anyway.

The thing is we all interpret the regs in different ways, i think most sparks just have to use a lot of common sense.