Yes, the current design is that every transaction entry is unique through its lifecycle.
If it were a requirement that virtually identical instruction sets should be differentiated and concurrently processed, we could e.g. include an account nonce in hashing, but I don't see the motivation for splitting approvals atm
_Originally posted by @s8sato in https://github.com/hyperledger-iroha/iroha/pull/5217#discussion_r1831052876_