I don't think that if a 51% attack happens, then the coins in addition to Bitcoin will live, immediately a panic will start in the market and everyone will sell Bitcoin, because no mistakes should occur, and canceling transactions is quite the opposite of what Bitcoin was created for.
Of course, maybe in the future there will be a new coin that will be protected from such manipulations, but it will take at least 5 years to restore trust for people, I think so
51% is not unknown and the costs and complexity of executing it is how PoW solves the Byzantine general problem. If you think governments care about Bitcoin at all, to the extent that they're willing to spend billions on trying to harm the current Bitcoin, then you are wrong. You cannot prevent 51% attack but the opportunity costs that arises from this is immense. If it ever happens, then we roll back to before the attack and change the algorithm. By doing so, the government has effectively wasted billions of dollars for nothing.
51% attack is not a Bitcoin killer because the game theory behind it hasn't failed. The sheer energy required to execute the attack is equivalent to certain smaller countries. The land area required to house the ASICs and the manpower required to set it up would be fairly big as well. Without understanding the difficulty of executing an attack, and the benefits of doing so from the government's PoW, then you won't understand why this isn't a problem.