Correct, but can't you see the problem?
The goal isn't to have the team with the most wins or the least losses win. The goal is to have the best team win.

So if those teams are 1-1 against each other which one is better? One team won the first match, then lost in the winners bracket where arguably the stronger teams should be. The other team lost the first match then won against weaker opponents in the lower bracket and then won the second match. It's not obvious which team is actually better.
That's why using longer matches for later rounds makes sense. If you lose a Bo3, but win a Bo5 it would make more sense if you're the better team, statistically speaking. Especially if you win the Bo5 against higher seed advantage due to coming from the losers bracket.

By the same logic using a match of around twice the length for the final and no bracket reset also makes sense.

The basis for best-of matches is that the team that wins more maps is better. If you play a Bo3 of Bo3s you can win with less maps won. That's a direct contradiction to the assumption the format is based on.

So two matches for the grand final is obviously not the only fair choice because it isn't even fair. Playing a Bo3 in the upper bracket finals and a Bo7 in the grand finals is obviously fair from the perspective that the better team wins. But what advantage do you give the team that won the upper bracket final? A Bo5 also allows winning with a drawn map score. Bo4 is fairly reasonable in terms of overal map score but Stark complained about the "random map advantage". So no, the choice is not obvious.

Usually with long game times where everything else is Bo1 and Bo3 isn't feasible or where every game will be played on a different day playing two matches with a bracket reset makes more sense, but that's not the case here.