--snip--
If the block size limit is lowered, the cost per transaction increases. There is the potential that transaction fees in total will increase more than the block size limit will decrease. So lowering the limit would increase total transaction fee revenue.[...]
In this case the network's hashrate will increase and it will be harder to maintain the attack. If the avg fee goes above 1% then the block size limit will increase in the next epoch.
There is nothing in this scenario that would cause the network hashrate to increase on its own. You are describing this as an attack, however it is something that would benefit all miners. As such, there is little reason why all miners would not participate via a cartel (something similar to how OPEC works for the oil markets).
If there were a situation in which bitcoin must have a dynamic block size limit, it would be superior to have the block size limit based on a sat/vByte basis. If the cost to include 200 bytes in a block becomes too high, the block size limit will increase, and if the cost to include 200 bytes in a block is too low, the maximum block size will decrease.
The problem with this proposal is that we don't know the real value of the fee, if the price of btc is high then the fee will be too expensive, if the price of btc is low then the price per vByte will be too low, we can determine a percentage but we can't determine btc's price.
The price of transaction always has been, and always will be measured in terms of BTC. If the price of BTC is too high, the market will respond accordingly.
I would also point out that your 1% transaction fee target is very high, even by traditional banking system standards. The cost of sending a wire transfer is at most $15 or $25 (if not waived), but it would be very unusual for someone to send a $2500 wire transfer. It is far more common for someone to send a six-figure wire transfer. It might cost $0.20 to write a check, but it would be very unusual to write a check for $20.