feat: multi-user sessions + text/image chat with OpenAI and secure env loading

[teachingandlearning4077-prog]

Motivation

• Make the QuantumX chatbot fully functional for multiple simultaneous visitors with per-user histories and a richer feature set.
• Allow advanced ChatGPT-like behavior and image generation when an OpenAI key is configured while keeping a robust local fallback when the API is not available.
• Avoid hardcoding secrets and provide a safe .env-backed configuration path.

Description

• Add app/settings.py with get_env() that reads environment variables first and falls back to a local .env file.
• Rewrite the engine into app/engine.py to return a ChatResult (text + optional image_base64) and implement OpenAI-backed text completions and image generation, plus local fallbacks for math, summaries and lists.
• Add multi-user session management in app/main.py using a qx_session cookie so each visitor gets an isolated ChatEngine instance and /chat accepts mode=text|image.
• Update frontend templates and static assets (app/templates/index.html, app/static/app.js, app/static/styles.css) to support mode selection and render generated images inline.
• Add tests/test_engine.py and tests/test_settings.py to cover the new ChatResult flow, image-mode fallback without an API key, and .env behavior, and update README.md, .env.example, .gitignore, and requirements.txt accordingly.

Testing

• Ran unit tests with python -m pytest -q and all tests passed (6 passed).
• Performed an automated import check python - <<'PY'\nimport fastapi\nPY which failed because fastapi is not installed in this environment, so the live web server could not be validated here.

---

Codex Task

[teachingandlearning4077-prog]

.