...
You are falsely conflating everyone switching ("nobody really cares about Bitcoin any more") with loss of marginal use cases (EG SatoshiDice and coffee). That's not how the substitution effect works. What actually happens those who gain the most value from Bitcoin stay and those who gain the least switch. That is basic economics.
When oil prices rise, we demand/produce/consume (IE substitute) more coal, nat gas, and alt energy instead. That doesn't mean nobody really cares about oil any more.
With a 1 MB blocksize limit one has 3 tps (reasonable practical estimate) for the current Bitcoin network. This translates into approximately 95 million transactions per year.
So here are some questions:
What non marginal use cases do you expect the average person to be able to use Bitcoin for?
Do you expect on average those transactions to be mined within that average person's life expectancy? That of their children? That of their grand children? That of their great grand children?
Please do the math.
At 3 tps it wouldn't be average people using it. With 10 billion people an average person could only make one transaction every 100 years. So that would have to be some sort of foundation for other systems (side chains, Lightning, etc.) that support higher volume.
BTW, Lightning doesn't work very well with small blocks, though maybe that can be fixed (in its present form, you need enough capacity to close many channels in a relatively short period of time). Sidechains probably don't either, although it seems at least possible that one or more side chains could become so popular that the main chain isn't used much by anyone except by miners (unlike Lightning you don't necessarily
need main chain tx capacity for a side chain to operate). 100 million tx/year might be enough for that.
This still ignores the question of how the main chain is secure and remains decentralized, how sidechains are secured, and probably other unanswered questions with the approach.