One problem is that there's lots of things to consider when you're reviewing a setup than just town:scum win balance. For example, excessive swing is less fun than an imbalance, often: if you make a setup where the scum can be effectively defeated by D2 and it actually happens, the scumteam will be very upset. Likewise, you have to ensure that the setup won't be decided by role assignment (if there's a role that badly needs to not be NKed, you don't want it to go to a player who always gets NKed N1, but there's no way to ensure that when randomizing roles correctly, and thus you have to get rid of the role altogether). You want to try to preserve the mod's idea behind the setup (unless it's utterly unworkable), so that it isn't just a case of the NRG picking out setups for mods. And of course, you need the setup to be one that'll actually be fun to play.
In general, this may mean that being mildly (but only mildly!) scumsided might actually be more fun than being balanced. The easiest way to "balance" a scumsided setup is to increase the chance of the town completely locking scum out early, but they won't enjoy it if that happens. In an extreme case, imagine a meta in which scum are generally very good and town are pretty bad by comparison (not saying that necessarily describes the meta here; it's a thought experiment). Pretty much the only way to get a 50:50 win rate will be to give town a chance of winning the game at random despite a lack of scumhunting ability, and the scum will feel highly cheated if it actually happens.
From another point of view, imagine a game that ends in 2:1 lylo. Then imagine a game which ends in 2:1 lylo but with a confirmed townie. Having seen these situations repeatedly from a number of viewpoints, the former feels more balanced to the players (regardless of alignment), but the latter is closer to balanced theoretically, and most likely in practice as well (it's very hard to read players in 3p lylo and looking back over the history of the game isn't as helpful as you'd think). Now consider what sort of games lead to these endings. Hitting 2:1 with no confirmed players is pretty easy. Hitting 2:1 with a confirmed townie, though, means (barring combinations involving bulletproof players who can either confirm players themselves or be confirmed by other players) there were two confirmed townies at the start of the night before. Two! Tweak the numbers a bit, and suddenly the setup-with-two-confirmed-townies-that-became-a-balanced-endgame is a setup-with-so-many-confirmed-townies-that-scumhunting-is-pointless,-town-win-easily-and-the-scum-are-upset.
And from another another point of view, 10:2 and 11:2 mountainous are both disastrously scumsided (in addition to their other known problems), and yet it's common for both factions to feel like they have a chance through much of the game. (I suspect 9:2 or maybe even 7:2 mountainous would actually be
more
enjoyable, despite being even more scumsided.) This probably works better if the players don't know it's scumsided, though.
And from yet another point of view, I once replaced into a game for which there was only one set of actions that gave my faction even a theoretical chance to win, all the actions in question would need to have been taken by other players, and they had no reason to do so in any case. I enjoyed myself, right up until the end when I realised it was impossible to win (but then the game was almost over so I just played it out). What was important was simply that I didn't know I had no chance to win, because it let me play to my win condition, scumhunt, drive lynches, and all the other things that are part of the game.
Anyway, after all that arguing to say that games being scumsided is not necessarily indicative of a problem, I will say that I'm not convinced that there isn't, in fact, a problem. Some of the most scumsided games could probably have been made more balanced or even townsided with changes that wouldn't have fundamentally changed the nature of the setup nor allowed for a blowout, and doing so would be a good idea if possible.
I think the most useful thing to do here would be to find specific examples of scumsided setups that have been allowed through the Normal queue, that a) were reviewed for balance; b) were perceived as scumsided postgame; c) (not required but would be helpful) have a public review thread. Then we can debate how it went wrong and what should have happened instead.
(As a counterpoint: did any excessively townsided setups slip through the review net recently? I know you sometimes get cases where players claim the setup was townsided, but oddly those setups are often won by scum, which means that the players may well be incorrect in their assessment.)