[skj]

Google crawlers come from well-known IPs, especially well-known to Google. Appengine requests come through reverse proxies, and there is no fundamental difficulty in not counting requests from crawlers towards the quota. That said, see my other descendant comment.