Fix label mapping, retrain model (99.26% accuracy), fix Confidence DNA display, add src/ config/ and training scripts

Update: Docker, FastAPI deployment and improved sentiment analysis

Revert "feat: replace hardcoded stocks with dynamic Wikipedia scraper (S&P 500 & Nifty 50)"

feat: replace hardcoded stocks with dynamic Wikipedia scraper (S&P 500 & Nifty 50)

fix: migrate to PyTorch and optimize for Render 512MB RAM limits

chore: update configuration for Hugging Face Spaces (port 7860 + metadata)

feat: complete Docker and FastAPI deployment configuration