I would think the race car analogy would be more an example of the detrimental effects of wake turbulence. It takes power to push the car through the air and that power, IIRC, is linear. In the case of race cars the following cars require less power for the same speed as the lead car does a disproportionate amount of the work of moving the air around the vehicle, setting up the wake in which others can follow. In the windmill example the blades use the wind energy to generate the lift to make them turn. Less total wind energy would be available to turn the second and third sets of blades due to losses to turbulence, rather like a competition yacht spoiling the wind of a competitor by turning in front of it.

What my brilliant possibility requires is either the ability to dither the wind direction back and forth by a few degrees or to rotate the array at a similar rate and I cannot imagine a feasible way to do that. I wrote before I thought.  

As the Dutch said while fighting the Spanish: "It is not necessary to have hope in order to persevere."

I just reviewed my basic aerodynamics.

Reynolds number is the ratio of inertial forces to viscous forces

Laminar flow has a Reynolds number below 2000

Turbulent flow usually is somewhere above that because thtere is a transition state, it is not exactly at 2000

The Coefficient of Lift is defined as the lift force over the inertial forces so if the inertial forces are high (as in turbulent flow), the Coefficient of Lift is larger

The Coefficient of Drag on the other hand, is defned as the forces of drag over the viscous forces, so a lower Reynolds number (in laminar flow) would have more viscous forces and a higher Coefficient of Drag.

Therefore, turbulent flow should provide a more efficient ratio of Lift to Drag.

Counter-intuitive? I would expect the second row, which is the first to encounter wake turbulence, to have the biggest drop and then for drops to increase slowly WRT the first row as one moves back in the array. The first row is the only one not to encounter wakes. What is considered the maximum acceptable drop allowed for the last row of an array? Any rule of thumb?

As the Dutch said while fighting the Spanish: "It is not necessary to have hope in order to persevere."