Good reading!
I think I understand the direction and you have started strong regarding knots and its challenges but then you just went on about core and bip110 and suddenly you said knots it is for me.
2 questions
A/ do you imply that arbitrary data on core and consequently bip110 is a bigger change/problem for you where as knots than seems to be the smaller “evil” ?
B/ did you dig further into the knots dev structure especially the single point of failure you have mentioned?
I am a node operator since some time and have updated mine to core v30 but with a limited op_return threshold, assuming this way I may navigate through this, but also lacking the time to really understand the issue at hand, however your article has helped, thanks