Reachzero
I feel that people have not really thought through "how is Salamence different from Lucario, Magmortar, etc.?" When we talk about "preparing for Salamence" it is qualitatively different from preparing for "obviously OU" threats like Lucario, Gyarados and Heatran. The reason is twofold.
1. Salamence is immediately threatening to a degree that most OU sweepers are not. If the immediate response to Salamence is wrong, the opposing player will likely lose a Pokemon immediately. This is especially true in the case of "MixMence" sets, that is, sets that have Draco Meteor. This puts heavy pressure on the opposing player to switch. The fact that Salamence has base 100 speed means that there is a very large pool of Pokemon that it threatens with a very low level of risk--and unlike, say, Jirachi, it is virtually impossible to wall. Switching against Salamence is almost a necessity (especially, though not exclusively, for stall and semi-stall teams), yet switching is never safe. Most importantly, the threat of MixMence gives Dragon Dance Salamence chances to stat up that it would likely not otherwise get--for example, Skarmory will almost always switch out the the first time it sees Salamence, since most players will not risk losing Skarmory to a Fire Blast (of course, the Salamence user can always choose to Fire Blast Skarmory later if he/she needs to anyway).
2. Salamence is incredibly difficult to revenge kill without specific priority moves.
Again, base 100 speed is a major factor, but the real back-breaker is Dragon Dance. MixMence is at least moderated in its effectiveness by the fact that it is fairly easy to revenge kill and can even be switched into with the help of a little prediction (though that is risky, of course). Once Salamence gets a Dragon Dance, however, the list of options becomes very short. The most common (and effective) option is to use Scizor or Mamoswine to revenge kill Salamence with their powerful priority moves. Yes, I realize that Weavile can do so as well, but Weavile has difficulties in the present metagame unrelated to Salamence (and did before Scizor obtained Bullet Punch....). While this is admittedly effective, both Pokemon are relatively easy to lure out and easily walled. Scizor in particular can be virtually 100% dealt with using Magnezone. Yes, Scizor does hurt Salamence's effectiveness somewhat, but not as much as, say, Blissey affects Latios' effectiveness. One Pokemon is not sufficient in itself to keep any other in a lower tier, particularly one that is relatively easy to remove. Of course, the second method of revenging Salamence would be to use a Choice Scarfed revenge killer, yet the list of potential OU revenge killers of over base 100 speed is short--Jolteon, Azelf, Starmie, Gengar, Latias, Infernape. Yes, some of these are commonly used with a Choice Scarf (Gengar and Latias, once in a while Azelf or Infernape), yet all have clear weaknesses as revenge killers, such as being easily walled and severely hurt by Pursuit. Choice Scarfed revenge killers tend to be a poor response for Dragon Dance Salamence. The third method is to use a bulky water such as Swampert or Suicune to survive a +1 Outrage and KO back with Ice Beam; the downside is that neither can be switched safely straight into Salamence (because of the possibility of Draco Meteor), yet if the bulky water is coming in to revenge kill, the player is sacrificing most of a second Pokemon's health just to kill Salamence, which seems to me somewhat excessive.
All of this is not to say that Salamence should necessarily be Uber. Each of the above methods works "enough" to not make Salamence an automatic win for its user. However, I believe that this provides enough evidence that Salamence is unusually difficult to deal with by OU standards that it merits status as a Suspect.