I think fundamentally you are looking at it a bit wrong. In theory, the aperture equivalence is exactly canceled out by the decrease in sensor area. When viewed at the same final size, the perception of noise should be about equal between your full-frame F/5.6 @ ISO 1600 and micro-four-thirds F/2.8 @ ISO 400.

But you note

I have rented the EM5 and notice the noise performance is about half a stop to a whole stop behind my 5D2.

This isn't due to an inherent advantage but simply a testament to the amazing tech in the OM-D E-M5, and the difference three or four years of sensor development makes. If you compare to the newer 5D Mk III, we'll be closer to the theoretical world, with a noise advantage of about two stops: the same as the crop factor, and therefore the same as the aperture difference.

There's a certain point in the real world, though, where performance is over the threshold in any case, and so while the smaller camera doesn't necessarily have any noise advantage, it doesn't have any meaningful disadvantage either, so you can consider other advantages (like size, weight, price, and so on).

On the other hand, you can shoot with a 50mm f/1.2 on the full-frame camera, and it's just not possible to get the equivalent 25mm f/0.6 in micro four thirds. And again, that might make a difference in the real world, or it might not. That's why we have options in the market, after all.