If the extractor area is the size of the bottle's cap (2cm x 2cm = 4e-4m^2) then to get a litre of water (1e3 m^3 air) requires 2.5e6m of air to flow past it - If you want it full in a day that's 30m/s... hmmm... doesn't seem likely.
However if the size of the extractor area is as large as the bottle (say 10cm x 10cm = 1e-2m^2) then a 1 m/s wind may do it.
Based on this I'm not confident in the "marathon runner" claim, but I'm sure techniques for extracting water from wind have been used before, but improvements could be made.
Scaling up the technology obviously increases the viability, yielding something like this: http://www.geek.com/articles/geek-cetera/prototype-wind-turb...
you are calculating the swept volume. but only a small fraction of the swept volume "makes contact" (ie is roughly within the mean free path of a water molecule) with the collecting surface (which appears to be necessary here for water to be extracted).
with netting your argument is closer to being correct. but this is an impervious bottle, not a net.
if you implemented the tube you suggest (and i understand that was simply an illustration to explain swept volume) then pressure would increase in the tube slowing flow and/or flow would be predominantly around the edges, with, again, a stagnant higher pressure region in the centre.
No idea how long that would take though.
[1] http://en.wikipedia.org/wiki/File:Relative_Humidity.png