If every Bitcoin user were also a self-custodial Lightning network participant, LN would fail due to liquidity routing issues.

This is a no brainer.

People with their 0.001 btc can't run a LN node that routes payments for those who need to transfer $1k. The whole network would end up with a giant set of nodes that can't really do much at all, and the increased number of nodes would result in making the routing problems even more apparent (those with less money in BTC are also those who are more hard up for money, and end up needing their own Bitcoin 'liquidity' more than whales or other LN rando participants).

I've never seen a small time LN node operators state they came out in profit with regard to operating fees (the fees they charge for making transactions via their node). I suspect even big nodes may operate at a loss at this point, although I freely admit I haven't looked at the data there.

Then there is the final nail in the LN coffin, which is that the more nodes, the more open/close channel txs you need, on a "settlement" layer that's already over capacity some of the time, and not looking to get capacity increases anytime soon.

Anyway, the point is, for these reasons, LN centralizes around big liquidity hubs, which will be the banks in future if they aren't already.

Bitcoin had none of these problems.

And unlike Lightning, it can scale as a payment system.