We will see by that time. I think the L1 bandwidth is fine for the purpose of SoV. At the end of the day, tx fees will be what market participants are willing and able to pay.

I see self custody as a spectrum, starting with people withdrawing their sats from exchanges and other centralized and government-controlled platforms. Game theory will provide efficient and secure options for everyday transactions on L2-L3. Bad actors will lose users and fees. Good actors will benefit with more users and better fees.

Not everyone will want or need to use L1 and high base layer fees will drive innovation toward Lightning and sidechains to solve scaling. I define scaling as providing Bitcoin for 8 billion people. In my view that scaling will happen primarily above L1. L1 fees will find their equilibrium in supply and demand considerations of the market participants. 

 Bad actors only get caught out if it it feasible to self custody. If the fees are so high that it's not possible then any custodian can make a killing by running a fractional reserve and bankrupt good actors. Bitcoin gets captured by the banks, just as if the block size was too large for non specialised people to run a node. 

If this happens I think Bitcoin will fail to dominate and we have an ecosystem of shitcoins 

 I’m wondering if there will be a way for proof of reserves to solve this. There might be a way to know say how much ecash has been issued and how much BTC is held in reserve.