here there is no corresponding valuable UTXO for each dust one?
It depends how you will define "valuable". Zero satoshi simply means "skip amount validation for old nodes". It can be treated as a valid, non-zero value on another chain. So, "valuable" depends entirely on the chain that is stored on Bitcoin. If it is valuable (on another chain), it will be spent. If not, then it will be left untouched.
you people are planning to increase the dust memory overhead of UTXOs to higher ratios than now
Sometimes you simply don't want to attach any value to some UTXO. Things are simpler and cheaper where you have zeros. It can be seen clearly on testnet, where you can just use zeros, use nonzero amounts to pay for fees, and assign any values in your software. Also, some UTXO's should have zeros, because for example if you have some NFT, then you want to assign zero, to show that no monetary value is attached into it and it is entirely separated, and there are no coins involved, you have just some token and some owner of that, determined by the Script.
Also, I have some other questions: imagine that zero satoshis (or just dust amounts) will be invalid (or distributed automatically to miners). Then:
1) How to represent things that should not have any monetary value? For example NFT's?
2) How to create fractional satoshis when they will be needed? You would have 0.493 satoshis, how to store that on-chain if it would be $493?
3) What amount should be put to outputs, where that value will be known later? For example, in decentralized mining you don't know how many coins should be received by each miner, when UTXO's are created. You know that after collecting all of them and getting enough confirmations.
Edit: 4) What if Bitcoin community will ever want to move to Monero-like system and hide coin amounts? Putting zero in that case seems to be a good choice. After your proposal, this change will be rejected.
If you want to get rid of dust UTXO's, then I think you should put an incentive for those who will do that. So, something like fees based on UTXO consumption and creation could work better than taking UTXO's by force. In this way, you can still have such outputs, but they will be more expensive (for example if you could pay 0.5 satoshis per byte for consuming a lot of dust inputs, then some people may start doing that).
