Just to get this out of the way, the counterargument "I want Feature X in the game -- if you don't like it, don't use it" never carries much weight in serious design discussions because it ignores the reality that if some feature is in the game, many people will use it.
In the first place, expecting people to voluntarily gimp themselves is just not going to happen in a competitive game. And in the second place, these are massively multiplayer games we're talking about. When a bunch of people use some feature, it can affect everybody in the game, including those who choose not to use that feature. It doesn't matter if I use Feature X or not; I'm still affected by it if it's in the game I'm playing with other people.
A persuasive argument for some feature in a game lays out specific, positive reasons why it would be good for that game. "Just don't use it" doesn't pass that test; it doesn't explain anything.
Now, that said, let me try to approach the characters-per-server question from this angle: What I'm trying to get at is not how a game should be played, but how it should be designed.
No one is talking about changing a game that's already launched (such as WoW) to reduce the number of alts allowed -- SWG already showed us what a bad idea big changes can be. The question here (as I understand it) is about what's best for games (such as Star Trek Online) that are still in development. It's not "how many alts can players have," but "how many alts should players have" that's of interest.
And for me, yes, it does come down to "alts unbalance a game's challenge level." If some game were designed so that every player's level of challenge was dynamically calculated based on how many alts they use and how frequently they swap stuff between those alts, I'd see no problem with that game allowing two or five or ten alts since the number of alts would be factored into maintaining an even playing field for all players. Likewise if a game defined its challenge level by assuming that most players would use all available alts.
(That would be a pretty strange design, BTW -- if the game's designed to be played with multiple alts, why not just design main characters to be more powerful/effective? The one way a multi-alt game might make sense would be if all alts were mutually exclusive in their abilities, so that "playing the game" actually requires swapping back and forth between alts. This has actually been done, though not in a multiplayer game. In the Infocom game "Suspended" you played as a cryogenically frozen mind who must control multiple robots -- each with very different capabilities -- to repair a planetary weather control system. It worked pretty brilliantly as a single-player game, but to my knowledge it's never been attempted as a multiplayer game.)
Most games don't dynamically calculate challenges based on alt usage, however. They hard-code a challenge level into hostile mobs, create zones containing mobs with challenge levels in a specific range, then expect players to decide which zones to visit. In that kind of game design, the challenge level is a static one-to-one comparison of the mob's level versus the (current) character's level... which means that using multiple alts makes every alt stronger than the individual character that the game's mobs were designed against. And the more alts available, the greater the imbalance.
So yes, it does seem clear to me that for a conventionally-designed MMORPG (one with mobs with static, character-based challenge levels), the obvious need is to minimize the number of alts. That's not the only solution; the game could also be designed to deliberately require the use of multiple alts, or to dynamically calculate challenge levels of mobs based on how many alts one operates.
Frankly, I'd find both those two latter types of games much more interesting than yet another conventional zone/level-based design. But it's unlikely that any major MMORPG will do follow either of those design paths. So for a conventionally-designed game, limiting the number of alts -- in order to keep everybody's level of challenge in line with the static difficulty level of mobs -- is the most effective option for that kind of game.
The fact that most MMORPGs allow alts despite being designed around static challenge systems doesn't mean those designs are "right," or that they're the best possible approaches to allowing players to experience more of a game's content.
Good designers -- of any kind of system -- don't allow themselves to be blinded to what might be created by seeing only what currently exists.