passing a common test suite (which we'll extract from go-ipfs)

@jbenet calls this out in #1. Judging by #8 people seem to agree with the principle that developing against a testable spec is a priority.

I'm curious how we're going to go about accomplishing and structuring this goal. I realize the actual written spec is still in progress (I myself am still coming up to speed on the protocol specifics).

I'd like to use this issue to center discussion on planning for how we're going to tackle this issue.