I was reading this discussion on adding nLockTime to Monero, and the OP says that we can achieve this functionality by having a small amount of cryptocurrency transferred with an transaction-unlock-time and use this as a lock by having a multiple input transaction with this locked account to create a transaction that won't be valid until some time in the future.
The question is are these multiple real input transactions binding? Meaning I have money in a 2-of-2 multisig account A where I own one of the keys and money in account L, where L is locked using transaction-unlock-time, and I have transaction that will be valid using A and L as real inputs back to to an output I hold (will be valid after L unlocks).
If I was a bad actor, could I separate the transaction and just send the money from A back to me by separating the transaction? Or is this a hard problem?
EDIT: The paragraph talking about multiple input is as follows: "It is technically possible to implement nLockTime now (without protocol change), but it is a little clunky. You have to put a negligible amount of Monero in a time lock transaction, and then sign a transaction spending it and the funds you want to nLockTime. This won't be publishable until the other transaction is unlocked, but it is a little clunky, requiring a couple of transactions (and data must be published to the blockchain)." The bold section implies a multiple input transaction.