First off, the argument is not about whether or not players will overuse Latias. It's about the inherent broken-ness of the suspect test.
Say I'm an average Shoddy battler. If I wanted to whore Latias, I would be on the Suspect Test ladder. However, if my team isn't packing Latias, I don't want to be on the Suspect Test ladder because the stats aren't permanent and there's one less threat to account for on the Standard ladder.
So the end result is that you get two kinds of people on the Suspect Test ladder: People who want to use Latias, and people who want to battle a couple times in order to be able to vote (and will probably use Latias anyway).
This creates a situation in which carrying multiple Latias counters is incredibly profitable to any team on the Suspect Test ladder. Where the likelihood is rather large that your opponent will be carrying stuff like Heatran/Lucario/Gyarados etc...
Counters to those Pokemon are rather wasted if the Pokemon they're made to counter isn't there.
However, given the astronomical likelihood of Latias being used, there's no reason not to have a team pack at least two Latias counters.
This is why we're seeing massive spikes in Tyranitar/Metagross/Blissey in the ladder.
So teams on the Suspect Test ladder turn out to be something stupid like Latias/Counter/Counter/(Pokemon that beats counters x3)
Or you go TAY's route. Dragon Dragon Dragon Steel Steel Steel.
Salamence Dragonite Latias Heatran Metagross Scizor. You have at least three immunities to ground, four resists to fire, three resists to water, one resist to rock, only two SR weaknesses, multiple high-powered physical and special attackers, and voila. Replace Nite with Flygon and you've got only 1 SR weak and an immunity to Electric and another resistance to Rock.
The point here is that this "overcentralization", as one may call it, is going to happen partly because Latias is good, but is inevitably going to happen simply due to the way the Suspect test is set up. There are two reasons to play suspect: to use Latias and to vote (and voters should be testing Latias out). Thus, the metagame has no choice but to be artificially centralized. Which means that the experience that is gained has no bearing on how Latias would fare in the standard ladder and metagame.
The experience will be worthless, the stats will be worthless, and the test will have almost no bearing on Latias' actual performance in the tier, which undermines the entire Suspect Test process.
Furthermore, we can't wait until Stage 3 to fix things. If too many fuckups occur, we won't even get to Stage 3 and the whole thing falls apart. At this rate (1 in 3, optimistically) it'll take a year to get there.
There are very few reasons to play suspect. Experience will be either terribly skewed or minimal.