I'm afraid I don't understand Bitcoin programming nearly enough to answer this.

But I think this is going to be a case of "buy a house on-chain, buy coffee over Lightning", where participants will weigh the risks to decide which layer to use. For instance, these "microtransactions" (another villified word) for API/emails will surely take place over L2/L3 but a large farmer might want to secure a price for his produce by timelocking funds on-chain, and binding them to an oracle-arbiter (a third party to discern whether goods have been delivered per contract).

This might also not happen at all. I can equally see a future where Bitcoin simply replaces US Treasuries as a backing asset, and even a unit of account, but the vast majority of it, at least in the corporate world, is centrally custodied by banks and there are proprietary networks that move it around much faster and cheaper (and only batch-settle, say, hourly).

And I think we're precisely going through a test period right now, and I hope it's spurring silent innovation, with builders trying to work around a problem (unfeasible fees for retail transactions) and we will see the fruits in the coming months/years.