Thanks for linking to my thread over on Donkey's forum. These videos are so interesting, what an ingenious method of visualizing flow dynamics! Guess I'll still copy/paste my post that you recommended I post over here (with some minor changes)-
The segment below immediately made me think of the P-Channel, and so potentially adding to the theory of why it works so well. Where gas of higher density is added into the flow at the top of a lower density gas.
Fresh air entering the burn tunnel through the P-Channel is presumably cooler than the air (flames) already in the burn tunnel. This causes an unstable inverted stratification which provides for turbulent mixing, as shown below. Although, in our situation we may want the air to be preheated in order to prevent excessive cooling of the burn and aid in efficient combustion, it won't ever be preheated to higher temperature than the combustion itself. Thus the effect still has potential, in theory. Whether or not this actually has a significant effect, I don't know.
I figure this concept in combination with Peter's idea of the volatile gasses tending to flow at the top of the burn tunnel, and these also coupled with the tripwire, makes for efficient mixing and a nice clean burn.