This is fantastic!

I had always thought the argument that large arbitrary data txs would inevitably be priced out by fees was a hopeful rationalization at best - but I nonetheless uncritically accepted it as plausible.... Until now: