Even at highest levels players make significant, but human, mistakes. On the other hand there exists a great amount of opening theory, with lines studied in depth, especially the mainlines. Isn't the precision of theory evaluation and human imprecision in contradiction? Is it generally difficult to punish small opening sins even at the highest (human) level (Evans-Gambit, Cozio defense, 1. ... a6)? If so, is the fact that most advanced players learn and play some kind of mainlines just some kind of gentlemen's agreement? Or do they score significantly better?
The score of openings is more often a combination of the strength of players which play it, the situations in which the openings are played, and of course the objective merit of the opening. Looking at the database you will often see sidelines which have much higher scores than mainlines, often these are trendy lines which are played by players who are up to date on theory and playing for a win against weaker opponents. That doesn't mean these lines are objectively better than main lines, most of the time main lines follow a traditional ~51% type of score, as strong players are often choosing them when playing to be solid.
For a specific example lets consider the Nimzo-Indian: 1.d4 Nf6 2.c4 e6 3.Nc3 Bb4. Here the three most popular moves 4.e3,Qc2,Nf3 all score around 52% (megabase), the sharp and trendy 4.f3 scores 53.5% and the old line 4.a3 scores a miserable 46%. These are all main lines and the only conclusion I would infer from these numbers is that strong players are playing 4.f3 for a win against weaker players. While black players in general find it easiest to win against 4.a3. If I was playing a player rated 400 points below me I would be much more tempted to play 4.f3 as I know the chances of them making a mistake is much higher than me making a mistake. However if I was playing a player 400 points above me, I would rather play a safer move like 4.Qc2, as there even if I make some small mistakes it will be easier to hold a draw.
To refer to the first part of your question, I am certain that all of these 4th moves in the Nimzo-Indian contain long theoretical lines which end in drawish positions. In a practical game between human players however it is more important who has a better understanding of the position, and luring your opponent into a position which you understand better than they do is one of the main goals of the opening. So in the end the score of an opening is largely irrelevant. Main lines are fundamentally sound positions which often afford players to make small mistakes without serious deteriation of the position.