Distributed query processing with S3 benchmark

[danthegoodman1]

Can we make a 1-10TB dataset in S3, merged up and optimized, and then run a distributed query across 8-10 32 core machines 👀

[danthegoodman1]

Discover with gossip, but maybe implement a "transport" interface for how it can listen for work, and how it can emit work to a known set of peers.

This is all optional and has to be enabled.

Maybe can use SidecarD project to handle cluster discovery and communication, but maybe we keep it isolation (that can be an impl maybe)