[Sherlock] Add AI-Driven Account Risk Assessment & Bot Detection

### Description Description This proposal introduces a new module to Sherlock for evaluating the authenticity and risk level of discovered accounts using multimodal AI (image + text analysis). While Sherlock excels at detecting username presence, it currently lacks risk scoring or bot/impostor detection. By combining screenshots, profile metadata, and deep learning, I aim to transform Sherlock into a more intelligent OSINT tool for cybersecurity professionals. Integration: Add a CLI flag (e.g., --risk) to trigger this module. Include risk data in the output file (JSON/CSV). Optionally visualize results in a UI/dashboard for review and export. Impact: Speeds up manual OSINT review by pre-filtering suspicious accounts. Enhances threat intelligence and bot/phishing detection. Helps investigators prioritize targets and automate triage. Request: I would like to further explore and prototype this feature, possibly as an optional extension or module. Please let me know if this direc