feat: async engine with aiohttp for ~2x performance improvement

[matthew6s]

Closes #2797

Summary

Adds an async engine using aiohttp as a drop-in replacement for the synchronous requests-futures ThreadPoolExecutor approach.

Benchmark Results (real-world, 478 sites)

Engine	Timeout	Time	Results Found
Sync (current)	15s	39.0s	13
Async (new)	15s	21.4s	14
Sync (current)	60s	65.8s	13
Async (new)	60s	63.6s	14

The async engine is ~1.8x faster and found an additional result that the sync engine missed (likely due to more efficient connection handling).

Changes

File	Change
sherlock_project/async_engine.py	New — async engine module
sherlock_project/sherlock.py	Import + CLI flags + call routing
pyproject.toml	Add aiohttp dependency

New CLI Flags

• --workers N, -w N — max concurrent requests (default: 100)
• --sync — use legacy synchronous engine

Backwards Compatibility

• Default behavior switches to async
• --sync preserves the existing behavior exactly
• Return value is identical (same dict structure, same QueryResult objects)
• All existing CLI flags work unchanged

New Dependency

• aiohttp ^3.9.0