Temporarily allow inconsistency where the risk is very low and/or limit access to a single partition for each UTXO (different UTXO limited to different partitions). Then full resolution at block confirmation time. Also by employing partitions interblock, the data that needs to be propagated on block announcement is radically reduced, thus the block period can be significantly reduced without driving up the orphan rate to unacceptable levels.

The thing was halting my programming lately is I wasn't satisfied that I had solved the permissionless and decentralization principles. But I think sending PoW with each transaction solves it, but I need to go over all the details again to make sure I haven't missed something.

If I feel I have the Bitcoin killer and something that can really help the world, then I will be motivated to program. I don't want to waste my effort. I am in extremely bad situation in my life and I can't afford to go down any nonproductive forks at this juncture of my life.

Edit: for me if there isn't a very viable chance of maintaining decentralization, then I am not motivated to code it. I feel with near certainty that Bitcoin (and other coins that employ Satoshi's PoW design, e.g. Monero) is headed towards centralization and 51+% control (even if the block size isn't increased). And for the reasons stated upthread, I view PoS (and DPOS) as non-viable direction to invest my effort long-term (they have served some transitionary role in crypto given Satoshi's design is also flawed).