Yea, sorry, I thought I deleted this post.
Quite frankly, I'm really not sure exactly how all this works, but I have a theory. My theory is that if everything isn't "just right", rather than giving me proper error messages, the test framework just moves on and then fails catastrophically, kind of like a compiler might do if somebody hadn't put a great deal of effort into proper messaging and recovery.
It seems that once I got everything "just right", it all works correctly. Well, mostly correctly, anyway. I still haven't figured out how to run the tests subsequent times without getting failures, so it's not perfect. But my initial run from "meteor run" all seems to work.
Thus, I tried to delete this post.
Apologies again.
David Logan