The gain is pedagogical: giving kids a good intuition about angles is so much easier when the constant you're working with represents an entire turn around the circle (360°) rather than a half-turn of 180°. The advantage of using tau instead of pi is much smaller in other situations, but when it comes to measuring angles in radians, it's huge. And kids who have a better understanding of angles and trigonometry are just a little bit more likely to become good engineers. So persuading math teachers that there's a better way to teach trig is an investment in the future whose potential payoff is 20-30 years (or more) down the road.
PI is to clever by half.
I'd really be curious to see any substantial proof for that claim.
The first time pupils encounter pi isn't when measuring angles. At least over here, that's still done in degrees, which is much easier to explain, and also latches onto common cultural practice (e.g. a turn of 180 degrees). So I suppose that already makes them good engineers.
But the first time pupils encounter pi is when computing the circumference and surface of a circle. While the former would look easier with the radius (tau * r), it looks just as weird when using diameter or when using it for the surface.