But seriously, who is paying for network security in the future when block rewards are 0 and all users are on L2?

I don't buy the settlement layer argument, that aggregated L2 transaction fees will somehow contribute to L1 fees paid to miners. All network users are incentivized to pay minimum fees possible. Bitcoin only works as a global money system if essentially all traffic can be handled on L2, which means fees on L1 will also be near 0. The 400,000 transactions bitcoin can support per day on L1 is far far too few to support global adoption without L2, but also much more than is needed if it is only used as a "settlement'" layer. This is the primary design flaw in the miner incentive, because there is essentially a cap in daily transactions the network can handle, either fees will be near 0 if transactions are below the cap, or they will begin to skyrocket as demand goes above the cap. If L2 solutions work, this means demand on L1 will be below the cap and fees will be near 0. If L2 doesn't work, well then Bitcoin can't have mass adoption. For bitcoin to be the gobal money standard, there will have to be multiple entities in the future willing to operate at a loss to secure the network, and a huge loss at that. They will need to be spending 10s of Billions a year to make sure the network isn't attacked. Who is that going to be?