Try loading blocks from a flat file

[teor2345]

Motivation

Zebra takes about 6 hours to sync from the network, but less than an hour to load blocks from a file on disk.

We could share a file of blocks using a faster distributed file sharing protocol, like Bittorrent or IPFS. Then sync the final few blocks as normal, using full validation.

For now, we just want to investigate how fast it would be.

Designs

Like copy-state, load blocks into the state. But instead of using a Zebra state as the source, use a flat file containing block data.