If segwit were to be a hardfork? What would it be?

Would it change how transaction IDs were computed, like elements alpha did? Doing so is conceptually simpler and might save 20 lines of code in the implementation... But it's undeployable: even as a hardfork-- it would break all software, web wallets, thin wallets, lite wallets, hardware wallets, block explorers-- it would break them completely, along with all presigned nlocktime transactions, all transactions in flight. It would add more than 20 lines of code in having to handle the flag day.  So while that design might be 'cleaner' conceptually the deployment would be so unclean as to be basically inconceivable. Functionally it would be no better, flexibility it would be no better.  No one has proposed doing this.

The position of a transaction in the blockchain should define which version of the rules is applicable to it

You could have a rule that you can refer to inputs using either txid or normalized-txid.  That maintains backwards compatibility.  The problem is that you need twice the lookup table size.  You need to store both the txid to transaction lookup and the n-txid to transaction lookup.

The rule could be changed so that transactions starting at version 2 using n-txid and version 1 transactions use txid.  This means that each transaction only needs 1 lookup entry depending on its version number.  If transaction 1 transactions cannot spend outputs from transaction 2 transactions, then the network will eventually update over time.  It is still a hard-fork though.