Segwit soft fork demonstrated by using soft forks you can always do rule-breaking changes (like change the transaction/block format) while still not being rejected by the old clients.
Non-standard =/= breaking rules. What rules are being broken? Give an example. And then explain why a network split does not result.
if a miner minted a block today that had 1
segwit-like TX in it, i seriously doubt that block would be valid.