> If the "abuse" in scraping is the number of requests... limit the number of requests?

Step one in doing this is identifying which requests are coming from the same agent. Guess what fingerprinting helps with.