Add Sambanova Inference

[seyeong-han]

Add Sambanova Inference

Ssmabanova Inference provides a free Llama inference server, allowing me to expand the selection of inference options for my project.

This pull request introduces a new integration with the Ssambanova service for LLM inference within the Llama Stack project. The major changes include the addition of configuration files, Docker and Conda setup, and the implementation of the Ssambanova inference adapter.

Integration with Ssambanova:

• distributions/ssambanova/build.yaml: Added a reference to the Ssambanova build template.
• distributions/ssambanova/compose.yaml: Configured Docker Compose services for Ssambanova.
• distributions/ssambanova/run.yaml: Defined the Ssambanova run configuration, including API endpoints and providers.

Implementation of Ssambanova Inference Adapter:

• llama_stack/providers/registry/inference.py: Added Ssambanova as a remote provider specification.
• llama_stack/providers/remote/inference/ssambanova/__init__.py: Implemented the adapter initialization for Ssambanova.
• llama_stack/providers/remote/inference/ssambanova/config.py: Defined the configuration schema for Ssambanova.
• llama_stack/providers/remote/inference/ssambanova/ssambanova.py: Implemented the Ssambanova inference adapter, including methods for chat completion and embeddings.

Documentation and Templates:

• llama_stack/templates/ssambanova/doc_template.md: Added documentation for Ssambanova distribution, including setup instructions.
• llama_stack/templates/ssambanova/ssambanova.py: Created the distribution template for Ssambanova.