Aggregate snapshots

[martinzima]

We need to be able to save and load state of event-sourced aggregates into a serialized state (e.g. JSON) for faster aggregate loading. This should mostly work as an optimization feature for mor demanding use cases a should be optional.

Thought: If we were saving serialized JSON state into database, we might also be able to use this state for querying the event-sourced aggregates even without their read model, given the DBS supports it (e.g. Postgres with its jsonb columns).