Enable using multiple LLM-compatible clients

[yodakoi]

Introduced a new environment variables:

• LLM_CLIENT_ENDPOINT → This allows dynamic selection of the API endpoint.
• LLM_CLIENT_MODAL → This allows dynamic selection of the LLM modal.
• LLM_CLIENT_API_KEY → Authentication key for LLM Client API.

=> Adjusted API call logic to use this variable, making the client behavior provider-agnostic.

• LLM_CLIENT_ENDPOINT: API endpoint for llm client

  - For OpenRouter, use https://openrouter.ai/api/v1
  - For Groq, use https://api.groq.com/openai/v1
  

• LLM_CLIENT_MODAL: Open IA Modal

  - For OpenRouter, use 'google/gemini-2.5-flash'
  - For Groq, use 'meta-llama/llama-4-scout-17b-16e-instruct'
  

• LLM_CLIENT_API_KEY: Authentication key for LLM Client API