V1 [OPS] WAL integrity check + repair tool

[austin-barrington]

Context

When the WAL becomes corrupted (unclean shutdown, disk error), the only recovery options are "restore from backup or delete the wal_dir" (docs/user-guide/troubleshooting.md:67). There is no hyperbytedb check or hyperbytedb repair subcommand to:

• Scan the WAL for corruption
• Verify RocksDB consistency
• Recover as much data as possible before discarding

Deliverables

hyperbytedb check subcommand

• Verify RocksDB WAL integrity (open, iterate all entries, check for read errors)
• Verify metadata store integrity
• Verify chDB data directory accessibility (list tables, check for obvious corruption)
• Report summary: entry count, last sequence, any errors found
• Exit code 0 if clean, non-zero if issues found

hyperbytedb repair subcommand

• Attempt to salvage readable entries from a corrupted WAL
• Rebuild metadata from surviving data where possible
• Create a backup of the corrupted directory before modifying
• Report what was recovered and what was lost

Checksums

Optionally (if WAL checksums are implemented per #XX [data checksums ticket]), the check tool should also verify payload checksums and report mismatches.

References

• docs/user-guide/troubleshooting.md:67 — "restore from backup or delete"
• src/main.rs:32-46 — current subcommands: serve, backup, restore
• RocksDB DB::Open() failure modes

Priority

P1 — Data integrity, production requirement