Use SQLite as our storage backend

[ghost]

Why

• We are starting to accumulate a few different things to persist on-disk, esp. as we will soon start working on Network resilient to node's crash-recovery failure mode #1079: Accumulated server outputs, head logic state, network messages and indices vector...
• Each fragment of state persisted is now store in some file, using a custom Persistence handle and logic
• Reliably writing to disk is notoriously difficult and error-prone
• SQLite has an unmatched track record of being a very versatile, reliable, lightweight, and efficient storage engine

What

• Replace custom Persistence logic and files with a single hydra.sqlite database containing tables for each type of item we persist

How

• Need to depend on sqlite-simple package or equivalent
• Replace or refactor Persistence handle: We could use a similar interface and hide sqlite storage behind it?
• Configure the DB location at node's startup