The first one -- the simple Park Tool gauge, is perfectly adequate for most uses. It provides two levels of indication, and is reasonably inexpensive and foolproof. The second one is maybe a little better, in that it gives you a sense of how close to the "break points" you are.
With either of the first two, and some other designs, one needs to be careful to not "force" the gauge -- too much force not only yields bad readings but also will damage the gauge.
With most gauges the chain needs to be measured while under tension. For a standard derailleur setup simply measuring the top chain segment while on the bike generally suffices, but for a fixie you will need to apply modest pressure to the pedal to make the chain taut (unless the tugs are tightened all the way to begin with), and if the chain is not mounted you will need to manually stretch it or apply a weight. The force does not need to be a lot -- just enough to press lubricant out from between pieces, etc.
The referenced article enormously overstates the potential error from roller wear. If you're measuring over 14 links like the Park tool, you accumulate 14 links' worth of "stretch" but only one link's worth of roller wear, so roller wear is a negligible contributor to the measurement.