reconfigure vector store size

[dayesouza]

This pull request introduces a patch to automatically synchronize the configured vector store size with the actual output dimensions of the selected embedding model. This helps prevent mismatches between the embedding model's output and the vector store configuration, improving reliability and reducing manual configuration errors.

Configuration validation and synchronization improvements:

• Added logic in validate_config_names (in validate_config.py) to detect the actual embedding dimension from the embedding model and update parameters.vector_store.vector_size and all index schemas if there is a mismatch. This ensures the vector store configuration always matches the embedding model output.
• Introduced a helper function _sync_vector_store_dimensions to encapsulate the synchronization logic, with appropriate logging and safety checks.
• Added a patch release note describing the new behavior. (.semversioner/next-release/patch-20260315024056229023.json)

Type checking and code clarity:

• Added a TYPE_CHECKING import for LLMEmbeddingResponse to improve type safety and clarity in the codebase.