Re: The Barry Silbert segwit2x agreement with >80% miner support.
It's tricky, because there has to be a method of preventing replay attacks. Obviously people would be critical of their efforts if they didn't ensure a clean fork. It's almost as though they're damned if they do and damned if they don't. And because the blocksize cap itself was rather a crude cludge to begin with, undoing it probably won't be any more graceful.
Not really. They chose to not use a hard fork bit in the header because they wanted to hijack all the SPV clients onto their fork. A HF bit would have avoided the need for their flaky activation mechanism but then they wouldn't have had their hijack mechanism.
NYC agreement is best and final. In 30 days this will be over without any chain split or problems. 100% smooth transition is planned. UASF will not happen. 
For the segwit part of segwit2x this might be true, but definitely not for the 2MB part which comes 3 months later. Even for the segwit component, if the miners lose faith in the code because of the showstopper issue just introduced, they may revolt and not deploy it on July 21st as planned (though no mining pool has hinted at that so far.)