I have to agree with the people who have pointed out that Mary Sue is a fan fiction issue, not an original story issue. There are expectations in a fan fiction that simply don't exist in original fiction. As was pointed out, one of these is that the cannon characters are expected be the leads. In original fiction, your characters are the leads.
Further, however, I think it is a bad idea to apply the tests for Mary Sue to original fiction. Why? Because it obscures real understanding of what makes a character good or bad.
A good character can't be constructed simply by avoiding 'they look like me' or 'everyone loves them'. You have to make characters who work with your unique story, and who have effective goals, and on and on. It's way harder than 'avoid Mary Sue', and there is already far too great a tendancy among unpublished writers to hunt for 'the easy trick' rather than do the hard work to learn what makes a good character and what fails.
In published fiction you can find classic characters who fit 'Mary Sue' (Dorian Gray was mentioned), and characters in wildly popular series fit as well. (Anita Blake was mentioned and I'd add Harry Potter). So, regardless of how you define success, you will likely find a 'Mary Sue' who succeeded there. This, to me, proves that a simple list of traits or the fact the characters appear to be wish-fulfillment are not enough to make a character bad.
In the same way, characters who have no Mary Sue traits can be complete failures. So, to me, the whole fuss about Mary Sue in original fiction is a waste of energy. Take the test for laughs if you want, but don't expect it to tell you if your character is good or not.