Allow incremental recovery

Currently state recovery logic operates in a way, where the entire journal has to be replayed before recovery succeeds. If this takes a long time and the process is aborted, it will restart from the beginning.

Expose a configuration knob, which will allow taking snapshots and purging journal entries in batches, i.e. for 10K journal entries we would process a 1000 of them and then create a snapshot and purge them.

This would allow the process to finish faster on subsequent attempts.