Parallelize lyrics fetching with ThreadPoolExecutor

[shimpeiws]

Summary

Currently cli.py fetches lyrics sequentially per song via client.get_lyrics(song). This works fine for small numbers of songs but becomes a significant bottleneck when using --max-songs 50 or higher.

Proposed Approach

• Use concurrent.futures.ThreadPoolExecutor (or asyncio + aiohttp) for concurrent lyrics retrieval
• Limit worker count to 3–5 to respect Genius API rate limits
• Ensure thread-safe access to the lyrics cache

Affected Files

• src/barscan/genius/client.py
• src/barscan/cli.py