Lightning (L2) scales transactions really well, but not users. Every self-custodial user of lightning needs his own channel(s). The idea behind virtual addresses under L1 addresses is to scale users too. Being able to prove virtual address balances with a BitVM rollup will allow a single transaction on L1 to represent a thousand transactions on L1.5, including channel opens/closes between L1.5 addresses.
Discussion
L2 is just beginning. I agree that Lightning is not easy to onboard the average person. My biggest criticism of Lightning is a requirement to run a node to self custody. Perhaps ARK can solve this. Liquid is also a scaling and feature solution although with a more specific use case. L2 and L3 are in infancy. eCash arguably scales users. We will get there. And I'm not sure Bitvm actually adds anything meaningful. It's early and I very well may be wrong but off the bat, bitvm seems like vaporware to me. Blockchains are not good databases. This is one of the main criticism of the EVM. That, and we don't need a different currency with a dynamic centralized monetary policy.
I agree that the self-custody onramp for Lightning is fairly advanced, too advanced for a newbie and definitely too much for grandma, which is partly due to it's early stage. To achieve self-custody for every average Joe and his grandma will be too much to ask with the existing paradigm. Instead, rollups are going to be necessary if automated/simplified LN nodes don't become dead simple.
As I said in another comment of mine on this thread, I would highly prefer a Bitcoin native solution over an "adapted" one. Whether BitVM is indeed that solution has yet to be seen, but I can be hopeful.