Add async parallel processing for batch operations

[ahmed-sekka]

Description

Implement parallel processing for batch operations to significantly improve performance on large document sets.

Current behavior

• Documents processed sequentially
• Single-threaded

Proposed behavior

# Process with 4 parallel workers
ragctl batch ./documents --workers 4 --output ./chunks/

Expected improvements

• 3-5x speedup on multi-core systems
• Better CPU utilization
• Configurable worker count

Tasks

• Add --workers / -j option (default: 1)
• Implement process pool or thread pool executor
• Handle errors gracefully per-worker
• Aggregate results correctly
• Show per-worker progress
• Add worker count to history/logs
• Benchmark and document performance gains

Technical considerations

• Use concurrent.futures.ProcessPoolExecutor for CPU-bound OCR
• Use ThreadPoolExecutor for I/O-bound operations
• Ensure thread-safe history writing
• Handle keyboard interrupt gracefully