I truly don't see much weight in the too few players per character argument, because that was essentially the case with the majority of the roster in Melee. There have been and there will always be underplayed characters. I don't see how citing Melee's simplicity even applies when it was achieved by essentially ignoring more than half of it's roster in competitive play. Characters in Project M may be ignored in much the same way, but changing design decisions to protect the current playerbase boundaries is a wasted effort, because the playerbase will inevitably change over time.
What makes matchups suddenly more important isn't that the roster is larger, it's that suddenly top tier characters don't get to power through a matchup with just one tactic. For example, a Sheik could be completely unfamiliar with the Roy matchup in Melee, but it wouldn't matter because chaingrab wins.
Given that the roster size isn't even the main cause of the matchup knowledge problem, I don't see a reason to limit it when so few slots are even being considered that they won't have an extreme impact either way.
If the argument is that top tier players really just can't be bothered to learn that one new matchup for a clone of an existing character, but would like to win anyways, I really don't know what to say.
Not to mention, it shouldn't be assumed that every single player will stick with one character and only be recognized for work with their main. When I look for good Roy players to learn from, plenty of them are actually playing Roy as a strong secondary. Trying to strike up a competition for best Roy main would be unnecessary and really not a concern that should change the design of the game.
I actually have high hopes that P:M will be able to give some weight to diverse players alongside the "main only" types. Fleshing out the playstyle spectrum seems like an attractive option to me because it allows multi-main players to have the option to make more subtle changes between characters, instead of "Here's gimmick 1, if that doesn't work I'll move on to gimmick 2." Instead, you are free to make small changes and carefully trade out strengths and weaknesses to best counteract your opponent's style while keeping an overall strategy that you are comfortable with.
I also don't see the point in asserting that adding any new characters at all is going to cause an unmanageable rise in complexity. Throwing in new niche or gimmick characters adds to complexity and excessive reliance on matchup knowledge, and I don't think that any more than 1 or 2 should be added for this reason, as well as the effort involved. On the other hand, a balanced clone that fits well within working knowledge of the game adds variety and less predictable gameplay. I would be just fine with the PMBR going to the slot limit with this kind of addition.
