My thoughts on this (and I am guessing here):
No insulation is perfect. There is always _some_ current leakage through the insulation, and thus a finite resistance.
In use, a small amount of current leakage must be tolerated between an energized conductor and its surroundings.
The quality of modern insulating plastics is so high that the leakage expected is far, far lower than one could actually tolerate.
The result that a resistance test which indicates what would be and _acceptable_ level of leakage (if evenly distributed) is _taken to mean_ that there is a defect in the insulation. The conductor might function just fine with 1 megohm resistance to surrounding material per 1000 feet of conductor...but insulating plastic is so good that if you saw a resistance so low (1 megohm per 1000 feet) you would rightly conclude that there is a hole or other defect in the insulation.
Because there is always going to be a small amount of current leakage, your insulation resistance test has to have some standard for the minimum permissible resistance.
It seems plausible to me that insulation would have _different_ current leakage values wet versus dry. Wet the insulation might cause enough leakage to _fail_ the test...but be good enough insulation to function acceptably. In this case, you might have insulation with no defect at all, and which would function acceptably when wet, but it would fail the insulation resistance test because of water.
On the other side of the coil, dry air is a rather good insulator. If you have an insulated conductor with holes in it, and the holes don't bring the conductor into contact with surrounding conductive material, then that hole would not cause a low resistance reading. This argues that if you actually want to detect holes in the insulation, you actually need to surround the insulation with a conductive medium, and perform your resistance test between the conductor and the 'conductive medium', eg. water.
-Jon
