My thoughts on the supply cap are basically that hodlers are free riders and so as long as that remains network security will trend downward after some threshold is reached. Haven't done the math on that threshold, so I couldn't give you an estimate of when.
I'm not the only person that thinks this, and in fact this was a well studied problem well before I came to understand it. Usually it's framed as "the miners need to be paid" and as a block size problem but actually it's more a problem with who pays and the incentives that emerge from that.
As far as layer 2s handling settlement, that could suffice, but the problem then is, a miner needs to be paid whether from a thousand small transactions or one big one. If everyone is settling on lightning or something those close transaction fees need to be large, which means channels need to be large, which is a centralizing force in lightning. We already see most users using lightning with large liquidity channel prividers, increased fees to open and close channels will make that problem worse, and would apply to any layer 2 architecture whereas the centralizing trend we see on lightning is specific to it's architecture.
I'm happy to see a die hard Bitcoiner willing to talk about these things, usually but not always I get dismissed when I bring this stuff up. It is a real problem that needs a solution if Bitcoin is to succeed.