fix: improve restore consistency and add fsck command

[KGFCH2]

@Rakshat28
Addresses #44.

Summary

This PR introduces major reliability and recovery improvements to bdstorage by hardening the restoration pipeline and adding a new integrity auditing command for long-term vault maintenance.

Key Changes

Atomic Restore Workflow

• Refactored the restoration pipeline to prioritize consistency and crash safety over batch performance.
• Database updates are now committed immediately after each successful file restoration and atomic rename operation.
• Ensures that unexpected interruptions (such as SIGKILL, crashes, or power loss) only impact the file currently being processed rather than corrupting the broader restore state.

fsck Subcommand

• Added a new fsck command for performing a full "ground-truth" audit of the vault.
• The command:
  • Walks the filesystem directly
  • Verifies hard links and stored metadata
  • Reconciles inconsistencies with the redb state
  • Cleans up leaked vault entries
  • Repairs incorrect reference counts caused by interrupted operations or prior failures

State API Extensions

• Added list_all_files to the internal state management layer.
• Enables efficient full-database traversal and integrity validation workflows required by fsck.

Result

These changes significantly improve the resilience of bdstorage against crashes and interrupted operations while providing users with a dedicated toolset for maintaining long-term vault integrity and deduplication correctness.