Even in a good test environment, with test cases and such, this is a very narrow time window.
Yeah, I agree with you there. They spend 14+ months putting this together, then want to spend maybe a month seeing how it'll work, with a very limited number of people testing it, most likely. But at least there is a somewhat public testnet and some advanced notice of testing. That rarely ever happens before hardforks around here.
What should one do, that likes only 50% of the changes?
Reject the hardfork. Make them scale it down to the changes that actually have consensus. Then discuss, test, and possibly revise the ones that do not have as much support. You know...collaboration and consensus stuff. Pretty novel idea, isn't it?
And as we know, there are many ways to reach consensus. Very doable.