Add support for persistence

[ezrasingh]

          The first step toward achieving HA might be to address fault tolerance by implementing persistence, as the current setup doesn’t support it—meaning a crash would result in total data loss. 

A possible solution could involve introducing state snapshots and an append-only log for operations. This would allow for some level of recovery during a failover, with configurable options for snapshot intervals.

My preference is for a memory-focused approach to persistence, where:

1. A snapshot of the current state is taken.
2. As operations are processed, they are logged.
3. When the log reaches a certain size or memory threshold, a new snapshot is taken.
4. The log file is either archived or simply discarded.
5. The log is then flushed, and the process repeats.

Originally posted by @ezrasingh in #5 (comment)