Recently I found an interesting testnet transaction: 952656e19ce57698880ad6ada935f29ee2cae82c4e7486516908b630671dd84b. It seems to be some kind of puzzle, but probably even the author cannot just move the coins without actually solving it
There is no puzzle here, there is only simple outputs that are clearly unspendable because they are breaking the consensus rule.
Each output in this transaction has something like this:
<8000000000000000000000000000000000000000> OP_LessThan
Script evaluation fails here simply because OP_LessThan pops an item from the stack (a byte array) and it expects it to be at most 4 bytes. The last item is clearly bigger so it fails.
(which is not the case in well-known mainnet transaction 08389f34c98c606322740c0be6a7125d9860bb8d5cb182c02f98461e5fa6cd15).
There is nothing special about this transaction. All of its outputs are the standard P2PKH script that we see every day, and the only way to spend them is to have the private key of those addresses.
OP_DUP OP_HASH160 <751e76e8199196d454941c45d1b3a323f1433bd6> OP_EqualVerify OP_CheckSig