Add server for consuming events, storing in DB, and recreating the tree.

[drewstone]

We will need a server implementation for rebuilding the merkle tree that functions as the relayer system similar to https://github.com/tornadocash/tornado-relayer.

This can be done by consuming the historical events of tree insertions and rebuilding the tree locally, likely in conjunction with a database of sorts. As the trees grows large, checkpointing will also be worthwhile.

On the topic of testing/rebuilding the tree, it might make sense to also support for the time being a flag that stores all leaves on-chain. Having the flag would reduce complexity on the client front, as users can just query the chain for the time being.