The reason the market consensus tolerates a limit to the left of Q*...
You have your hypothesis and I have mine. We will only be able to determine which is correct with the benefit of hindsight. Like I said upthread, the nice thing is that these hypotheses are at least partly testable. If your theory is correct (that the market will tolerate a limit to the left of Q*) then that would have the affect of pushing aggregate fees above the aggregate cost of production for block space. If the total miner fees collected over a six month period in the future were, for example, twice the total block rewards lost due to orphaning, then I think I would agree that your theory is correct.
Time will tell.
My hypothesis is supported by current observations of the network dynamics.
Yours is based on faulty assumptions derived from your "paper" that pretend "cost of production" (which I presume is another name for orphan costs) are non-negligible on the long-term and that it is not trivial for miners to overcome them by cooperating/centralizing.
By all account you are wrong as miners have already shown a tendency to optimize for profit. An unbounded block size will only grow this incentive.
Miners will mine the coin with the most value.
Only Mathematics rules, and no, it does not encompass any democratic procedures within it...