...
It only requires a very short duration to send the difficulty permanently into the stratosphere. Only a fork can revert it. The attacker doesn't have any ongoing cost. The rental cost is cheap. All your CAPEX arguments fail because you forgot about renting.
...
Incorrect. This is mounting a 51% attack for approximately one day in order to double the blocktime for approximately two days. I do find it interesting though when people suggest attacks that are way more expensive than a 51% attack and have relatively insignificant impact.
That's not how it works these days. Modern difficulty adjustment algos have sanity checks to prevent, or blunt the impact of, that class of attacks.
Why don't you get a blog, spend some time to structure your ideas properly and so on ? 