fix: improve model router error handling and add debug logging

fix: handle provider/model format in router (strip prefix before calling provider)

test: comprehensive AI provider testing with 90% success rate

test: add end-to-end scraper test script with successful validation

feat: update model router with NVIDIA support and latest model scores

feat: add NVIDIA provider and update Gemini/Groq with latest models

test: add comprehensive API and core module tests

fix: add list_providers method and agents list endpoint

fix: replace deprecated datetime.utcnow with timezone-aware

feat: add multi-model LLM router with providers