You're not missing anything, John. You can damage most Lionel multiple-output transformers that way. Lionel seems to have been oblivious to the problem. The only transformer with a warning against it was the KW.
Many electrically naive persons think that there is no problem with doing that, and especially if the outputs are from the same transformer. In fact, doing it with the same transformer is the most dangerous way, since the resulting fault current does not flow through the circuit breaker and will therefore not trip it.
I have often encountered the fallacious "we've always done it that way" argument, that no harm has ever come from it. This belief probably arises from the fact that small voltage differences and brief cross-connections from running from block to block do not draw enough current for long enough to burn anything up. That doesn't make it safe, since it is inevitible that sooner or later, some train will stall over the gap and not be noticed. Anyone who, unlike you, does not recognize the problem, will probably never know why his transformer burned up "for no good reason".
Neil Besougloff wrote an article in the July CTT recommending this very way of operating. I wrote a letter to CTT warning of the danger. I received a reply from Carl Swanson, the new editor, expressing skepticism and asking for more information, which I sent to him. He said that he would have a frequent contributor who is an electrical engineer "run some tests".
I had heard no more, when the next, September, issue appeared with an article by Ray Plummer showing the same problematic scheme. I wrote Carl again and attached a photograph of a damaged transformer showing a distinct burned section in the middle of the secondary winding, which I believe could have resulted only from the kind of fault I was warning against.
I also described my own tests: I measured a fault current of 11 amperes between the outputs of type-Z transformer set to a 5-volt difference; and 30 amperes with a 10-volt difference.
Carl replied that his engineer "hooked up an old-style Lionel transformer and measured the power flow through the device in every wiring combination he could think of." His conclusion was "that a fault condition will uniformly raise voltage across the windings, but at no time did the voltage approach the safe working limit of the windings." I can make no sense of this and can only imagine that there was a communication failure.
He went on to claim that Lionel transformers cannot be damaged by wiring faults. I know that this is not so.
He then invoked the "we've always done it that way" dogma and cited the experience of someone with eight transformers and block signals as evidence.
Finally, he asserted that in new transformers "[t]he waveform is always the same from the output of each 'handle'. There is no potential difference (voltage) between multiple controls regardless of their respective settings." I mention this irrelevant item only to show the extent of his electrical knowledge.
Carl was unfailingly cordial through all of this. I believe that he is completely sincere in thinking that there is no problem. I suspect that he thinks that I am a crackpot, although he would never be so impolite as to say so. However, he is an editor, not an engineer. I am posting this in the hope that other engineers and those with other electrical credentials will back me up by writing to Carl. From his point of view, it is his engineer's opinion against mine; and he knows the other guy.