Very interesting read!
My impression is that you have found some interesting ideas to borrow from Koinos, which is good!
I agree with you that in the future, depending on the congestion of the network, you probably will require hundreds of mana in Koinos in order to make some transactions, and in this case people will have to buy mana delegations from others, so like paying a fee for the tx.
I think this is normal. Nothing wrong with that, if the network is really congested you will need something to prioritize your tx over the others, like paying for it.
The community can also decide through governance to increase the hardware requirements in the resources contract (no hardfork required) in order to reduce congestion and then reduce the mana requirements for users. However, this comes with the cost of lower decentralization because less miners will be able to reach this hardware requirements.
There is one topic that you didn't explore and I think is very important when designing economic incentives for spam mitigation:
It's true that eventually in koinos the exchanges will drive the mana market. However, on EOS this is driven by stakers. So the stakers have profits for mining (pay per vote) and for using the network resources (pay for use). This could end up in a wrong economic model like the famous Eidos contract on EOS, where there is no incentive to reduce spam. On Koinos is different, you have to select between profits for network resources (mana delegation) or profits for mining (burn your koins to get virtual hash power). So here you have an incentive to not spam: which is take profits by mining.
I like your article. Looking forward how you will design the economic model of Fractally.