0 (or consider this a non-vote at this point)
The wikisearch example is a very difficult project to determine ownership over. The knowledge of the implementation and the (subtle) caveats is primarily outside the knowledge of the current list of contributors, like Chris stated. Meaning, it doesn't make sense to keep it underneath the Accumulo trunk as those who fix it have to operate proxied through other committers. The wikisearch code already suffers from rot, due to less direct love than it deserves/needs. Saying that we need to reduce complexity of the code doesn't mean anything if no one can actually work on improving things.
Although, I agree with Eric that the wikisearch code is an excellent example of the features, performance, and application of Accumulo which would make sense to keep with Accumulo to ensure that a real-life application is tied with each Accumulo release. For those who want to do something "serious" with Accumulo, the wikisearch can (and I would argue, should) be an excellent place to begin to learn about the intricacies and best-practices. There's far too much contained in functioning code that documentation can never encompass.
Something needs to change to keep the wikisearch code at the level it should be, but I'm undecided if the medium of hosting will really matter in the long run. Ill-maintained code will suffer no matter where it's kept, but change needs to happen before the example is moot.