What you describe is not even a +51% attack, it is a demand for a forced hard fork from one miner telling millions of other people what software to upgrade to. That is simply the most absurd thing I've ever heard. And the reason you think this will happen is because if that one miner claims "this change is actually necessary and good for bitcoin because the network is in grave danger", that would be accepted by the general public?
It is not "one miner". It is 51% of the haspower having decided to do something that will bring millions additional revenue to all miners.
And that is precisely why the 51% attack is a problem: because 51% of the hashpower may be just one miner, and yet that miner WILL prevail against all the others. That is a fundamental feature of the protocol. When it was designed, it was called "one CPU, one vote instead of one man, one vote", and it was believed that the difference between those two concepts would not be significant. Now we have seen that the two are vastly different.
What would happen in your example is the millions of bitcoin participants would tell that one miner to take a hike and continue with the existing P2P core protocol. Then when that miner started to produce nonconforming 25BTC blocks, they would simply be rejected out of the P2P network as invalid, and the honest network would carry on. The attacking miner could have 95% of hash power and it wouldn't matter, his blocks would be rejected as nonconforming and the network would simply re-adjust to the 5% remaining. You do not even understand the basics of how a potential 51% attack is limited in it's range of attack vectors in the real world and how those limited attack vectors are harmless/manageable.
As I have pointed out several times, users can reject blocks, but cannot create them; only miners can. The users do not own the mining equipment, the miners (obviously) do. So the millions of users can scream and cuss as much as they want, but they cannot control or harm the miners, not even deny them the block rewards. Even if all the users reject all the blocks that the miners mine, they will not be able to interact with other users, or get any transactions in the blockchain, if the miners will not serve them. And a cartel that has 51% of the hashpower can close the network to all competing miners, and to all millions of users who are using the current protocol, whatever that is. Even if the users try to mine with their CPUs, they will not have enough hashpower to get through the cartel's jamming. Is any of these statements wrong?
When the banks get together and decide to raise some fee, their millions of users may scream and cuss as much as they want, but cannot stop the banks from doing that. Some rebel users may take their cash out, and perhaps try to create their own informal banks; but most users would continue using the banks, because they would rather pay the extra fee than do without their services. That is reality. Why do you think that bitcoin users and bitcoin miners would behave differently?
You then claim miners and users who do not upgrade would have their transactions blocked. Again this is insane. What would happen is those miners and users would confirm transactions on their own honest chain (the original chain) and not care whether their transactions confirm on the attacker's invalid chain.
But that is not how a 51% cartel can jam the honest chain. Read my outline, PLEASE.