It shouldn't be that surprising. With every class having a different focus, especially when it comes to combat... any single type or style of fight will have a couple classes be head-and-shoulders better than the rest.
When its 1 PC vs 1 PC out in the open with nothing to hide behind, the fighter or barbarian has an advantage. When its 1 PC vs 1 PC from a hidden position with lots of cover, the rogue or ranger will. When its 1 PC vs a group of PCs, the wizard and other area of effect casters are always better as the single combatant.
But this is really nothing than can or should be "corrected" for the game itself. Because you'd have to change how every class performs in combat so that no one playing field, number of players, or style of fight gives anyone any advantage against each other. But that turns everyone into carbon copies of each other.
That may be true of D&D (for some reason I'm not clear about,) but I don't accept that as a universal truth when it comes to game or rpg design. I don't accept that because it's not at all true in other rpgs I play; even some of the rpgs I play which have a much wider variety of character types possible than D&D. Certainly, in those games, the situation, terrain, and other factors may give one particular 'build' (for a lack of better words) an advantage in specific situation or specific group of situations, but that same build might very well struggle in different circumstances. Other games I've played have pulled off (and done it very well) the idea that PCs are stronger when together and fill different niches while still having less disparity when the PCs come to blows. Though, to be fair, those games aren't D&D; D&D isn't one of those games, and the design considerations between D&D and the games I have in mind are drastically different. Still, my point is that it can be done, and I believe that having a concept such as bounded accuracy makes it more possible; not less. Then again, 5th's meaning of bounded accuracy isn't exactly what I thought I meant by the term when it was first mentioned. Though, even that doesn't change that it's possible to have a team game which still allows for characters to feel more even when faced with each other. I don't suggest D&D should do that; I merely suggest that doing it is possible, and I expected the current edition to be closer to that rather than further away from it.
I'm also of the opinion that PvP has never been a 'playstyle' that any edition of D&D has supported well. That does not mean it could not and was not done, of course. But I don't think it has been a major design goal of D&D in the past, nor should it be in the future. As others have said, it would constrain the design to much, and would perhaps lead to a 4e type 'everyone-feels-the-same' type of design that many rail against (whether true or not).
Now having said that, I don't want to be dismissive of the original post or the PvP play style. Such problems revealed here could also manifest in similar play styles (i.e. combat heavy) and could be useful in highlighting potential flaws in the system. I'm curious what the details of play session where. From the mentions of bounded accuracy I surmise that it was a fighter-type that 'went rogue' (no pun intended) and the rest of the party had a difficult time dealing with this? I would like to know what the classes in question were and which were perceived as weak in this scenario. Was it a simple 'surprise, I'm whipping out my sword and hacking you in two' situation? Inquiring minds want to know.
Basically, this is how the final encounter went...
The party got to the encounter with the Ice Witch. One of the barbarians in the party felt that some of the people from the towns should be made to suffer and didn't particularly feel the Witch's plan should be stopped. He cut a deal to join her side.
I don't know exact numbers of his character, but I can say he was a mix of Barbarian and Fighter, and focused on two-weapon fighting. His style was basically to throw out as many attacks as he could and then add the rage bonus to all of them.
On the 'good' side; trying to stop the Witch, was my character (halfing rogue/monk mentioned in other threads,) a barbarian, a wizard, a fighter, and a cleric. The wizard -not surprisingly- went down very quickly. What was more surprising was that the regular barbarian also went down rather easily when engaged in a toe-to-toe fight with the traitor. While the regular barbarian was dealing far more damage when he hit, he was getting less attacks. The traitor, between having advantage via rage and a multitude of attacks from TWF had multiple rolls with which to fish for criticals; he also easily made up for his single attacks doing less damage by being able to stack his rage damage bonus on all of the attacks. After dropping both the wizard and the barbarian, the traitor then moved onto my character. The only thing which saved me was that I had a much higher AC (17) than some of the other party members in spite of the fact that I was not wearing armor due to having a high dex and a respectable wis. Really though, even with that, it was just poor rolling on the part of the traitor. I could, to some extend, match the amount of attacks he was putting out, but his damage was higher, and he was still raging and such got advantage on his attacks. In the end, I won out, but just barely, and I contribute it more to his poor rolling toward the end of the session more than anything else.
Had his rolling been marginally better, he would have very likely dropped my character. Then, at that was left was a cleric who was ill suited for melee, and the fighter who had already been badly wounded earlier in the fight. One PC would have very likely killed the rest of the party. While I understand that PvP is not a design consideration for D&D, something about that scenario still seems wrong to me. A lot of things about that scenario still seem wrong to me, and I don't believe it all boils down to "well, that guy wasn't playing D&D right."