[C2] Extend support for multiple AI providers

[ywarnier]

Current C2 implementation provides support for OpenAI but the other providers (for which we already have a recommended settings structure) are not yet available to use.
Add support for these before the stable release.

[christianbeeznest]

Consolidated Chamilo 2 AI helpers across the platform: added/standardized multi-provider selection (per feature/type), exposed capabilities and provider lists from the backend, and wired the UI to use the selected provider while keeping existing flows intact (glossary/LP generation, quizzes grading, media generation, and document analysis). Also finalized the recommended ai_providers JSON settings structure (OpenAI/DeepSeek/Mistral/Grok/Gemini) and synchronized the settings value template without overwriting existing configured values. Current availability depends on provider type support (e.g., PDF/document_process may still be limited compared to TXT/text providers).

Thanks in advance for your review.

[ywarnier]

Maybe you can integrate Nano Banana in the Gemini section with this info?

                                                                                                                                                                                                                                                                                                                                                                                                                                     
  'image' => [                                                                                                                                                                                                                                                                                                                                                                                                                       
      'url' => 'https://generativelanguage.googleapis.com/v1beta/models/%s:generateContent',                                                                                                                                                                                                                                                                                                                                         
      'model' => 'gemini-3.1-flash-image-preview',                                                                                                                                                                                                                                                                                                                                                                                   
      'n' => 1,                                                                                                                                                                                                                                                                                                                                                                                                                      
  ],                                                                                                                                                                                                                                                                                                                                                                                                                                 
                                                                                                                                                                                                                                                                                                                                                                                                                                     
Important caveat: The request body format is also different from Imagen. You need responseModalities: ["IMAGE", "TEXT"] in the generation config to tell it to return images. So whichever PHP service class handles the image call for the gemini provider would need to branch based on whether the model is an imagen-* (uses :predict + instances[]) or a gemini-* (uses :generateContent + responseModalities).               
                                                                                                                                                                                                                                                                                                                                                                                                                                     
  You may want to add a discriminator field to the config to make that branching explicit:                                                                                                                                                                                                                                                                                                                                           

  'image' => [
      'url' => 'https://generativelanguage.googleapis.com/v1beta/models/%s:generateContent',
      'model' => 'gemini-3.1-flash-image-preview',
      'request_format' => 'generateContent', // vs 'predict' for Imagen
      'n' => 1,
  ],

  Sources:
  - https://openrouter.ai/google/gemini-3.1-flash-image-preview
  - https://ai.google.dev/gemini-api/docs/models/gemini-3.1-flash-image-preview