>There is limited evidence that farming use of antibiotics is what drives resistance in clinical cases.

This is a classic example of asymptotic behaviour analysis.

>That is - we use enormous amounts of antibiotics in farming

You're entirely correct, the data fits an almost exponential curve, the only limiting factor so far has been soil health and costs. the amount of biocides we have to use to keep our productivity will be impossible to match and therefore, even with 0 transference of resistance to human antibiotics, you're gonna see more transference to humans because of farming-biocide resistance. ofcourse, most of this will result in lower crop yield instead of actually selling bad product, this can be slowed down with essentially the same policies and techniques as hospitals employ. But the core problem is the exact same, the extremely quick and deadly takeover of a single strain resistent against our current quo pro due to overuse of a single line of defence.