Followed by rotational inertia by about a hundredth. Weight matters, obviously we are all quite obsessional about it, and it matters more on steeper hills, but it's nearly irrelevant whether it's at the rim or on your bum.
Imho that biketechreview analysis is not particularly scientific - many conclusions seem to be based on the small difference between two much larger readings from a power meter, which accuracy is not great - some might say it is a common schoolboy mistake...
It is not too difficult to work out exactly the theoretical difference in power requirement of accelerating a wheel with the same mass concentrating at the rim vs one concentrated at the axle. The former has inertia, and the wheel would require exactly twice the power to accelerate compared to one with the mass concentrated at just the axle (and therefore has no inertia). The calculation is shown
here. Conceptually we can visualise why it is so by appreciating rim mass at the bottom of the rim also has to be moved backwards when the wheel is moving forward, while the mass at the rim at the front of the wheel has to be moved downwards as well as forward etc. etc.
While it is true that a heavier wheel returns the energy when you decelerate, it is not much good to you if that energy happens to be the heat generated on your brake pads.
It is however indisputable that for most purposes the "speed performance" effect of wheel lightness alone is not great for most people. Nevertheless all else being equal if you have a choice you are better off having a light rim/tyre/tube than a light hub/frame/potbelly, and there are
other reasons for having good, light wheels.