I am confused. In your example, it is in the blockchain and since you have the ability to spend it, then why would any fork make it so you cant spend it?
The spending transaction isn't in the block chain.
You create transaction A and then create the refund transaction B. B is signed by both parties. A is submitted to the blockchain. B has a locktime of 2 years in the future.
A soft fork happens that makes B unspendable for some reason. Perhaps, it requires signatures signed with the original private keys. In that case, it is impossible for either party to create the new spending transaction.
This has already happened with the P2SH fork. If you happened to create a P2SH output, then it would be unspendable. On the plus side, I assume they actually checked that there were no such outputs when the fork was proposed.
The key point is that a (chain of) timelocked transactions that are spendable now, should also be spendable in the future.
I see, this was before CLTV, where future locktime tx couldnt be confirmed.
Theoretically any unspent multisig output could be in this state, and any p2sh output could also have this issue.
But why does the fork have to make the existing outputs unspendable? I know it is possible to make any sort of fork, but who is proposing anything that would make these locktime tx unspendable?