In bitcoin the UTXOs represent units of bitcoin itself, they are not called or considered "tokens".
If you use zero satoshis, then you can easily turn them into tokens, because then they can be combined in any way you want, and there is no limit to the number of UTXOs that you can create. Imagine some attack vector, where some altcoin does not have its own chain, but uses Bitcoin as a storage. You can put all coins into the UTXO database, or even flood it with non-monetary use cases, like domain names, and for example trace ownership by tracing a particular UTXO. Also, you don't need any satoshis, the only thing that is needed, is someone willing to put such transactions into the chain, because for any transaction, you can always attach some zero satoshi outputs, and it will still be valid. That means, even if all coins will be mined, and the basic block reward will be zero, then still, there will always be some incentive, because some people may still want to notarize things on Bitcoin.
It is always an output of a transaction
You can always attach some zero satoshi outputs to any transaction, and it will still be valid. Basically, testnet3 is just a decentralized token creation platform by definition, because all coins are defined as "worthless", so you cannot trade them. That means, the only thing you can do, is just collecting some satoshis to pay for notarization, and then publish some information on-chain. Because coins are worthless, and blocks are far from being full, you can always pay the minimum price of one testnet satoshi per virtual byte. And then, your balance just describes, how many bytes you can notarize on-chain, because it is the only thing you can do, according to the rules.
So, I still wonder, why people use Bitcoin as their notarization platform, when they could use testnet3, that is cheaper, allows non-standard transactions, allows zero satoshi outputs (which means unlimited token creation), and can be mined locally (by going offline for 20 minutes, the difficulty drops into the minimum, and then you can mine any test blocks on your CPU to see, if something is possible or not).
That's more of an attack surface you are describing involving copycat-coins which is prevented by adding "replay protection".
The biggest mistake of those copycat-coins is "replay protection". If they wanted to include transactions faster, then they could provide just that, without splitting the coins. In this way, it would be possible to go back to Bitcoin, in case of altcoin failure, and then users could have all altcoin features, while still having 1:1 peg with Bitcoin. Also, all altcoin transactions would eventually be confirmed by Bitcoin, just it would take more time, but still, it would be possible.
Also, that case clearly shows us, that it is possible to make some altcoin, attach it to Bitcoin, and then make UTXOs that will move on both chains simultaneously. Then, it can be used to deploy any sidechains, without activating BIPs like 300 or 301. Then, peg-in would be just a proof of ownership for a particular UTXO, pushed on the sidechain, and peg-out could be realized just by moving the coins on Bitcoin, that would also trigger the same thing on the sidechain.
They are actually enforced by a centralized set of rules created and decided by the creators of these silly things.
Then, why they push anything on-chain? There is no reason to do so. All that is needed, is just signing a message, and putting that signature on such centralized website, then it could be easily assigned to the first person who did that. And then, if all rules are not enforced by the Bitcoin network at all, they could be fully executed on that centralized website, without any need to touch any on-chain coins, or push any large OP_NOPs on-chain.
There already exists token platform creation altcoins that people should use to create such things.
True. Also, there are cheaper networks, like testnet3, where "pushing data on-chain" is the only valid use case, because all coins should be treated as worthless by definition. The only reason why testnet needs any coins, is just as a flood protection, in practice people could cover all test cases with zero coins.