As Bitcoin wouldn't survive many years with huge annual inflation rate (because everyone want to dump such money) - the same is valid for zero (or: almost zero) inflation rate (everyone want to hoard such money). That's why I'm very pesimistic regarding: "it's also possible that fees could just spike"...

The root of problem is that we have no interconnector between two separate worlds, digital Bitcoin and the real one.

I know such mechanism is rather not possible in Bitcoin, but in LTE radio network there is something called "open loop and closed loop", to quickly react and regulate radio power parameters in the air channel.

and the similar idea is to swap: difficulty adjustment towards constant coin supply (i.e. regulating difficulty to keep a block time preconfigured by developer)

for: block reward adjustment towards constant security behind the network, in terms of purchasing power (i.e. regulating a block reward to keep difficulty configured let say once per year by developers)

i.e. setting (i.e. keeping) constant security behind the network, in terms of purchasing power (in cross-sectional prices of gold, Big Mac, recent iPhone, etc) - this is an open loop

algorithmic mechanism with negative feedback loop executes that order from developers by constant (smooth) regulation of block reward - this is an closed loop

This way you could have stable, transactional money, with stable security behind, with stable equilibrium between two groups of opposite interests (miners and stakeholders), and with embedded option to smoothly correct the overall direction the network goes (also in case of emergency situations/unexpected events like disruptive hashing hardware appearance, global economic turmoil etc)

Perfect, but even less probable than swapping halvings for Milton Friedman's k-percent rule :)