feat(vertexai): support meta's llama models

[CorieW]

Support Meta's Llama models in VertexAI.

Testing

[gemini-code-assist]

Code Review

This pull request introduces support for Meta models (e.g., Llama) on Vertex AI by implementing an OpenAI-compatible API client and necessary model options. The changes include adding chat completion and streaming support, defining the Meta model configuration schema, and updating the plugin to resolve and list these models. The review feedback highlights the importance of ensuring tool call IDs and tool response references are present for API compatibility, and suggests adding support for resolving 'gs://' URIs for image inputs, as the endpoint may not support them natively.

[CorieW]

/gemini review

[gemini-code-assist]

Code Review

This pull request introduces support for Meta models (Llama) within the Vertex AI plugin. It adds a new VertexAiMetaOptions class for model configuration, implements request/response mapping for the OpenAI-compatible endpoints used by Meta models on Vertex AI, and updates the VertexAiPluginImpl to discover and resolve these models. Feedback includes an optimization to cache the API client to reduce overhead and suggestions to use GenkitException with specific status codes instead of UnimplementedError for better error handling consistency.