On the issue of the validity of ZTS measurements, I'm with SilverFox.
Yes, it can give misleading results. However, it would seem prudent not to mix batteries that it gives different readings for. In spite of any debates about how the ZTS works (and whether this is valid), I think it's fairly safe to say that if it consistently gives different readings for 2 batteries, then there is something different about those two batteries. What seems less reliable is the idea that if it consistently gives 2 identical readings, then this must mean the two batteries are identical. However, the question then becomes, "is there a better way" to identify whether 2 batteries are identical. To my knowledge, there isn't a method that has been generally accepted as being better, and I certainly can't think of one. I think the only certain way of finding out how much life a battery has in it, is to run it to death.
It could of course turn out that your results reflect the reality very well, e.g. perhaps your "20%" batteries won't last very long at all. If you have a single-cell light you feel comfortable with testing them in, the results would be interesting...