for sustained power output of any duration, pedaling a fixed gear crank is about as good as it gets for getting power out of a human body. See Bicycling Science by David Wilson, MIT Press, for some interesting quantitative analysis. As an example to address your specific suggestion of rowing, compare a rowing scull with the Decavitator (MIT built pedal powered boat).
You should probably correct "skull" to "scull." Reading "rowing skull" then immediately mis-reading "Decavitator" as "Decapitator" was an interesting WTF?!? on my part ;-)
The huge speed difference with a single scull probably has more to do with inefficiencies in 'the drive train' (moving your weight around in a rowing shell introduces dipping, which loses energy; using oars is less efficient than using a propeller) than with inefficiencies of the rowing motion vs a cycling motion.
http://rowingbike.com/site/EN/ shows a bike operated using a rowing motion. That machine is competitive with 'normal' recumbent bikes, except in climbs. The main parts also could be fitted on that Decavitator. I would expect similar speeds.
Interesting point - it may be that rowing is comparable to biking in power, but the ease of building a cycling drivetrain generally wins out for projects like this, because the parts are readily available in any bike store (the cockpit of the copter here is an off-the-shelf bike frame and crankset attached to the spools).
Deadlifting is excellent for getting maximum force, but not power (powerlifting is misnamed!). The highest power lifts - the Olympic, explosive ones - aren't well suited to tapping to run a drivetrain.
Cycling is, among many things, very smooth, which helps when dealing with such delicate machines.
I would guess it is a pretty good compromise between smooth, efficient and powerful (a smoother motion presumably allows a less robust drive train). They pretty much need all the watts the pilot can supply just to get off the ground.
Edit: Corrected typo noted below - "skull" to "scull" - oops.