Modeling AI inference with KubeVela and OAM

A local-first POC that models inference workloads with KubeVela Components, Traits, Policies, and WorkflowSteps. Results from k3d and Ollama, and how OAM complements Kubernetes inference stacks such as llm-d.