From your examples, France is really the only attack I can see that both must be done and that I would consider game ending under most circumstances.
From that alone we can say that some portion of games are going to be lost on dice alone. It’s just the nature of probability.
That said, most people who think they lose to dice simply don’t understand dice. Yes, you lost a battle you “should have won”, since you had a 90% chance of winning, and it cost you the game. But if you fight a half dozen critical battles at 90% that will cost you the game if lost in a particular game, you’re going to lose half the time outright.
I generally think people who go into France light are insane for this reason - sure, you’ve still got 90% chance of winning but that means you’re going to lose 1 in 10 of your games on the first battle. That’s insane. But most people don’t consider that a “real” loss, despite the fact balancing minimizing your risk vs reward of spreading yourself out is literally the entire point of the game.
But this is also why I don’t think the other battles are strictly game enders if they go poorly. When the board is against you you simply up your tolerance for risk. A 10% chance of winning a Hail Mary that will decide the game now is entirely preferable to losing with certainty in a few more turns. A 90% chance of in a game deciding battle is entirely too low if everything is going your way.