In my experience, if a coil of wire is conducting a current, there will be a difference of potential between the wires in different layers of that coil, because any length of wire will have resistance. This is why a coil can short circuit to itself- a turn to turn short. In my industry, a turn-to turn short is one of the more difficult faults to catch, and is usually caused by overhandling the wire during coil winding. Vacuum impregnation of potting varnish helps, but if two wires are in contact with each other, and their varnish coating is degraded, it will prove to you quickly that there is a definately a voltage between the layers of a coil.
Perhaps I'm not understanding a subtle difference between a magneto and an electric motor.
This is a very reasonable question. In the context of your question, there is no difference between an electric motor and a magneto (although, it actually is a transformer). However, this particular test is different than the one you normally would run on an electric motor. In the test I ran there is no current being conducted by the wire.
First, assume all the insulation on the wires themselves are fine, and a layer of insulation thick enough to withstand 2.5 kV was applied to the armature before the primary was wound on it. That is, assume the coil was properly wound. Since none of the leads from the coil are attached to the body of the armature during this test, applying 2.5 kV between the slip ring (which in turn is connected to the HT lead) and the armature body will cause a tiny current to flow for a moment until the potential of all of the windings are elevated to 2.5 kV with respect to the armature. Since all of the wires in the coil are connected together, once this potential is established there is 0 V between any two layers of the coil. Therefore, even if the insulation were missing from some of the internal coils, and two adjacent coils were shorted together, no current would flow between them -- in this test -- because there is no voltage difference to drive the current.
What may be confusing you is this isn't quite the test you would normally do for an electric motor. In an electric motor you would actively drive current through the coils by connecting your ohmmeter to the leads. In this case, current would
flow between coils that were shorted together.
What the test I did checks is if the various insulators between the coil and the armature body are good. It says nothing about whether there are internal shorts between the coils which, as you say, is something of great concern. To test for internal shorts requires a dynamic tester, like the Eisemannn or Merc-o-tronic. If there were internal shorts, but not ones so bad as to make the coil totally inoperable, that would be revealed by needing to apply a larger AC current to the primary. You can't do this dynamic test for shorts on an electric motor because it only has a "primary," so you can't measure the AC current required to create a spark from the "secondary."
Although it gets a bit off topic, for testing for internal shorts on a dynamo (which is completely equivalent to your electric motor), I have an Evoibi coil tester. It measures the losses in the dynamo at two frequencies. Since the frequency dependence of the losses from internal shorts is different from the basic coil itself, this reveals whether there are any such shorted turns. By the way, these testers were developed for testing electric motors, where the market is a bit bigger for diagnostic equipment than it is for old motorcycle dynamos. Unfortunately, the large core losses of a magneto's armature makes it impossible to use such testers to determine if there are shorted coils in them. When working on motorcycle electrics, one can't have too many testers...