well you go treat that paper as your bible, and ill use my own research and stats and experience as the formation of what i will continue to go by.

by the way, that paper has lots of if's and assumptions too. so take it with a grain of salt.
just because someone puts a fancy title page and calls it a 'paper' doesnt mean its well researched

transactions having a fee does not guarantee acceptance in a block
transactions having a fee+x does not guarantee acceptance in a block before a transaction thats just fee-x
pools are free to choose a transaction to include for any reason they please. so fe's and blocksize are not things the network should mess with because it doesnt help/coerce a pool into doing what a network wants.

again a pool could develop its own fee rationale and cause users to adapt.
EG pools could stop accepting segwit transactions unless they pay the same price a legacy tx pays
EG pools could stop accepting legacy transactions priced at a 4x segwit comparative and only accept legacy tx's with cheap fees at a 1:1 comparative

pools can decide to ignore certain transaction formats like LN CLTV's

you and your papers assumptions that pools automatically will grab the highest tx and aim to fill blocks is the biggest empty assumption of all