Harvest from S3

Issue Summary

I was wondering if it would be possible to consider a harvest configuration that creates and links nodes to files harvested directly from an S3 account?

While using S3 to reduce bandwidth costs is a great idea, I am still utilising a lot of bandwidth by processing the files through my server first. Being a South African based server, there is limited bandwidth available and high overage costs that I would like to see if I can reduce.

Thanks