fix: load .env file in call_llm.py and support provider URL fallback - #202

Summary of Changes

• Load .env automatically: utils/call_llm.py is documented in README.md as the verification script to run (python utils/call_llm.py) after configuring credentials in .env. However, load_dotenv was previously only invoked in main.py, causing direct execution of python utils/call_llm.py to fail with ValueError: LLM_PROVIDER environment variable is required. This PR adds load_dotenv(find_dotenv() or ...) at script initialization so it works whether run from the project root or the utils/ directory.
• Support <provider>_URL: Added fallback check for f"{provider}_URL" in addition to f"{provider}_BASE_URL" to match the example naming in README.md.
• Resilient genai import: Wrapped from google import genai in a safe import check so environments using other providers (e.g., Ollama) do not fail at top-level import if google-genai is not installed.

Testing

• Verified python utils/call_llm.py runs successfully and queries the configured provider from both the project root and utils/.