Arguing that any significant number of miners will indefinitely mine at 0 income is disingenuous.
It isn't only not disingenuous, it is based on real world empirical observation, which always trumps theories. For example, when Monero did our planned hardfork to upgrade to v2 blocks, there significant hashrate that didn't upgrade and continued mining worthless blocks for days if not weeks. AEON's planned hard fork was similar (though with less advance notice and perhaps less well-publicized). I didn't really pay attention to when the v1 fork stopped growing on either coin (or has it ever?). This was not a spontaneous attack catching-unaware miners by surprise either, it was an upgrade that was announced and well-publicized for weeks to months.
Since you have now downgraded (or one might say upgraded, in terms of plausibility) to 10x attacks, that only requires 20 minutes to start to produce blocks at the pre-attack hashrate. IF 90% of the hash rate abandons mining, then it would still only take 200 minutes. Either way there will still be high-fee transactions stacked to mine, blocks being produced with larger block sizes, and the hash rate will start to drop.
On the topic of the tail reward, the eventual reward is about 1/20 of the current reward. So at 20x the current price you would expect a similar hash rate. That only brings us to about 600 million market USD market cap (factoring in increased supply). At 1 billion USD or more the hash rate would be higher.
The difference between the 51% attack and the 10X burst attack is not that the total hashrate expended by the attacker is less, but that it is 10X less income for the other miners.
This is wrong. In a 51% attack other miners get no income since their blocks are rejected.
What ever you do to adjust the algorithm to compensate in one way, will open a vulnerability in another way.
No. Note that I didn't say attacks could be completely prevented. I said resistance could be improved.