You make a good point, nice catch. Technically we don't know what the fastest possible algorithm is yet.

Just to be sure I understand, doesn't the conjecture postulate that O(n log n) is the lower bound, so _anything_ beneath n log n would violate it?

Yes (technically the lower bound would be the small-o(n log n)). The conjecture is that it’s possible to do in O(n log n) time, which this paper proves, and that it’s not possible to do any faster (which is still an open problem).