This pull request introduces improvements to provider and model configuration logic in flow files and the flows_service.py backend. The main focus is on making provider-specific fields more dynamic, updating how embedding nodes are handled, and ensuring models are enabled in Langflow before updating components. Additionally, several flow JSON files are updated to remove hardcoded default values and disable loading from the database for certain fields.

Provider/model configuration improvements:

• Added new methods to src/services/flows_service.py for finding nodes by display name, determining node provider, and mapping provider names for display, which help streamline provider-specific logic.
• Enhanced _update_provider_components to handle multiple embedding nodes, dynamically select which nodes to update based on the number of configured providers, and update existing nodes for the specified provider first.
• Refactored _update_component_fields to dynamically update provider-specific fields, ensure models are enabled in Langflow before updating, and use a mapping for field values instead of hardcoded logic.

Flow JSON file changes:

• Removed hardcoded values (e.g., "WATSONX_PROJECT_ID", "OPENAI_API_KEY") and set them to empty strings in flows/ingestion_flow.json, flows/openrag_agent.json, flows/openrag_nudges.json, and flows/openrag_url_mcp.json. [1] [2] [3] [4] [5]
• Disabled loading from the database for fields like project_id and api_key in the same flow JSON files. [1] [2] [3] [4] [5]

These changes make the provider/model configuration more robust, reduce reliance on hardcoded values, and improve maintainability for future provider additions.