As several responses demonstrate, these considerations can be long and complex.

One simple answer is that not everyone will have enough savings to compete with the fees on L1. The bandwidth for a secure and decentralized store of value is limited and cannot be otherwise.

In a good case scenario we might achieve 7000 tx's per block. That's a million tx per day (7000 x 144) and 368 million tx per year.

For a population of 8 billion people, the above estimated bandwidth could be compared with a 21 year waiting time per transaction (if it was a queue system, which it is not). Since it's a fee system, some deposit amounts will just not be meaningful on L1 in relation to the competitive fees.

A more plausible figure is 4500 tx per block, equalling 648k tx per day and 236,682,000 tx per year. This gives a bandwidth of 33 years waiting time per tx if it was a queue system. (Again, which it is not)

Even if the bandwidth on L1 was significantly improved, not every person out of 8 billion would be able to compete with the fees. Some people can afford to use L1 hundreds, or thousands of times so the above calculations are merely hints that highlight the issue.

The solution *must* involve L2 - L3's.

It can't be 100% solved on L1 and this is not even related to Bitcoin but would apply to any L1 blockchain. L1 fees will outprice some people - this is inevitable.

Personally I'm happy to use Lightning and Liquid for amounts below $100. A free market competition on L2 - L3's combined with self interest (tx fee income for relaying) will keep the options secure enough.

As the L2 - L3 development continues there will be more options and greater security.

The potentially offensive part of this subject is that realism (A is A) is often misconstrued as non-inclusiveness. Reality is what it is and no other decentralized competitor to bitcoin can solve the problem for L1 bandwidth either. L1 has its benefits and drawbacks. 

 Fees are likely to be very high, but that's okay. Arks, ecash fedimints, liquid, etc will give us lots of optionality, lightning will bridge all the layers, & maybe signature aggregation will help most people retain the ability to have an onchain footprint of some sort.

https://fountain.fm/episode/cLNMTcWrFO3hlQXk8e1K 

 Absolutely.

Yet in my humble opinion, signature aggregation seems to be far more complex than to just keep small amounts on L2 - L3's where they can be spent easily.

Using competitive banking services is fine in my view as long as we don't store large amounts there and as long as nobody can expand the total supply.

It seems we assume that everyone wants to store $10 on L1 when other options for that are better.