perf: Optimize RAM footprint to stay under 512MB for low-memory containers

fix: Add safe response handling and error alerts for text and voice queries in web UI

feat: Optimize runtime thread limits and memory footprint

feat: Add Qdrant Cloud API key authentication & cluster connection support

feat: Add Qdrant vector database engine adapter with dual FAISS/Qdrant backend support

feat: Configure Groq Cloud API key and high-speed gpt-oss-120b model

feat: Replace Gemini with Groq Cloud Llama-3.3 ultra-fast LLM API integration

feat: Add VisionQuest Indic RAG core services, vector search, safety guardrails & web UI

Initial commit: Hacker House Goa 2026 Voice Indic RAG