So I was curious about the math being flung around here, and decided to have a look at it.
Using mathematica, and a standard angled throw equation, with some guesses, we can have a pretty good guess as to what's going on here.
First, I assume Mr. Incredible is 2 meters tall...He's probably taller, but the error of margin is negligible...
I assume he throws at about 30 degrees (that's pi/6 in radians) with -9.8 as gravity.
Solving for when the coconut is at the ground gives us a closed margin of where, and how far things are.
We can see it should take approx three seconds, throwing the coconut at 100km/h (27m/s) (Mr Incredible is apparently over 1500 times stronger than a man who can lift 100kg, benching 155tons (I assume imperial tonnes, ie 2200lbs), he could have likely vaporized the guy if he really threw the coconut...but I'm getting carried away)
Anyways, I watched the scene and in super scientific, clicking youtube videos repeatable for peer review fashion, I see it took 1 second for the coconut to arrive, Which means the solider guy was 10.95 metres above the ground.
Your initial guess was, 20m/s^2 ( I assume you actually meant 20m/s^2
Solving for that gives:
Giving us an acceleration at -22m/s^2
So, yeah, he definitely fell pretty fast. I won't bother solving for the girl, though, in reality she didn't fall 2 meters, she fell probably less than a metre, landing on high heels off balance and falling to the ground. (The speed of things falling on a curve can get quite fast! which probably makes it seem like she fell faster than it should be.)
Math aside however, I couldn't say why that particular scene seems to be so quick, compared to others, and if they're using some kind of rendering/physics engines, like in video games, and simulators, they wouldn't change the acceleration of things just for single scenes, if it wasn't intentional.
This concluding, they intentionally increased the acceleration for some kind of effect, as in other comments pointed out, changing physics for a specific kind of effect can make it more dramatic, or suspenseful, or even funny.
It wasn't a funny scene, nor a dramatic one (in comparison to the girl falling, which would also explain the changing of her falls acceleration), but it was an sneaking around action scene, changing the accel. here makes no contextual sense.
Thus it could be quite likely be a mistake or more likely to me. They just wanted it over quickly. Rendering time costs money.
The OP was correct in their comment in pointing out that the model of the coconut throw, assumes correct gravity, but a screwy fall. Which points further to the fact there is an intentional change in physics.
To delve further into the conspiracy, I have modelled friction, into the equation to find out if this is playing a roll of any kind.
The orange curve being with drag, blue without. Assuming a weigh of 600 grams and a coconut 300mm in diametre.
This barely changed anything. a new height at 1 second of 9.84213 metres.
This reduces the accel. to -19.6843m/s^2. This with a correctly modelled coconut throw, giving us an incorrect fall.
So I applied the same to a model with drag, and at ~19 m/s2
Here we see quite a big difference between our two systems and how far the guard fell. Taking the maximum height of 4.84445m and solving for the accel. Gives what I expected:
A complete reversal of physics, and the expected acceleration. Uff, that's not good.
Lets see how fast the guard actually fell, if they fell at ~-19m/s^2.
Oh..0.7 Seconds. Seems about right, considering my youtube clicking.
The OP seems to have simply stumbled onto the fact that gravity within the entire movie is actually twice that of reality in all scenes. Nice catch!