As much as I am a weight weenie, it seems there is something to the flywheel effect of a heavy wheel, as long as there is not so much climbing that the disadvantages outweigh the advantages. As evidence, I cite Ondřej Sosenka, who used 3.2 kg rear wheel to break the hour record in 05'. I have tried to find more info on this setup, but most references on the web only mention it in passing.
Also, there was that spoke weight device that was making the rounds several years back. A guy in Boulder had developed it and Taylor Phinney tested it as a junior if memory serves. It had weights that slid on the spokes, and had some sort of springs that would hold them in toward the hub, until centrifugal force overcame the springs and moved them outward. This was supposed to offer the benefits of the flywheel effect at higher speed, when they weights would ride outboard, but reduce the rotational inertia at low speeds for climbing. Obviously total weight would remain the same, so you would still have to drag that mass uphill, just with lower rotational inertia.
I have heard a few explanations for the flywheel effect being helpful, and the commonality of them all is that they suggest that it is advantageous to store energy in one condition in order to receive it back in another condition. This has been suggested to help, as Causidicus suggested above, during the dead spots in the pedal stroke, helping to reduce the micro decelerations that would occur twice per pedal stroke. I have also seen it suggested that the flywheel effect helps on a slightly longer time scale. We have all had the experience of the "zone"/strength or whatever you want to call "feeling good" coming and going within a single effort. One minute you are ripping like Cancellara, and then you start to feel less good for a bit, only to have your good form return a minute later. The reasons don't really matter, as the point is that it is suggested that they heavier wheel will help to carry you through those weak period with more speed. Last, but related to the above two, it has been suggested that the flywheel will help on SHORT climbs, as it will store energy on the flat leading up the climb, and help carry you up the hill at a higher speed. Obviously, if the hill is long enough that you slow to your standard climbing speed, then the wheel will become dead weight, but for shorter hills, the wheel can carry you up and over, and then you can "recharge" the wheel on the following flat or descent.
From my understanding, the main reason that this is advantageous, rather than being a zero sum situation, is that the wheel will store and redistribute your power output to points in the effort when you are going slower, thereby reducing losses to aerodynamic drag, while boosting overall speed for the effort as a whole.
Back to the OP's original question regarding freewheels though, I had forgotten to mention this (unscientific) spin down test I saw a while ago on those crazy Gokiso hubs vs DT240s (http://cyclingtips.com.au/2014/02/gokis ... bs-review/
). My interpretation of the chart is that freewheel drag is a great equalizer, at least between these two hubs, because as soon as it comes into play (the "chain on" condition), they are fairly closely matched. Also, within 240s results, it looks like freewheel drag adds (very roughly) about 20% to the drag when compared to the "chain off" condition. This was less of an increase than I anticipated, but on further reflection, I guess it may be due to the increased number of axle bearings engaged in the "chain off" condition, which partially offsets the drag of the freewheel mechanism itself.