feat: incremental + parallel indexing for large monorepos

[NeelPrime]

Summary

Implements incremental indexing (#5) and parallel indexing (#6) for large monorepos.

Changes

Incremental Indexing (default behavior)

• Compares file content hash before parsing — unchanged files are skipped
• Deleted files are automatically removed from the database
• Second run on same codebase completes in milliseconds

$ arbiterx map          # First run: indexes 300 files (3s)
$ arbiterx map          # Second run: skips 300 unchanged files (0.05s)
$ arbiterx map --force  # Force full re-index

Parallel Indexing (opt-in)

• Uses ProcessPoolExecutor for multi-core parsing
• Each worker gets its own TreeSitterParser (no shared state)
• Single writer merges results into SQLite (no locking issues)
• Only activates for 100+ files (overhead threshold)

$ arbiterx map --workers 8    # Use 8 cores
$ arbiterx map --workers 0    # Auto-detect CPU cores

CLI Flags

• --workers / -w — Number of parallel workers (default: 1)
• --force / -f — Re-index all files ignoring cache

Testing

• 109 tests passing
• Lint clean
• Tested with multi-project monorepo (7 languages, 22 files)
• Verified incremental skip + force re-index + parallel parsing

Closes #5, Closes #6