feat(mlflow): add GPU node scheduling pattern

Parent epic: #166
Integration branch: feat/mlflow-enterprise-patterns

Add GPU scheduling support for ML/AI workloads with node selectors, tolerations, and nvidia.com/gpu resource limits.

Scope

• Add gpu section to charts/mlflow/values.yaml: enabled: false, nodeSelector, tolerations, resources.limits["nvidia.com/gpu"]
• Update charts/mlflow/templates/deployment.yaml to conditionally inject:
  • nodeSelector from .Values.gpu.nodeSelector when gpu.enabled
  • tolerations from .Values.gpu.tolerations when gpu.enabled
  • resources.limits including nvidia.com/gpu when gpu.enabled
• Add comments explaining the pattern for vendors adapting it to AMD ROCm or other GPU providers

Files touched

• applications/mlflow/charts/mlflow/values.yaml (add gpu section)
• applications/mlflow/charts/mlflow/templates/deployment.yaml (conditional GPU blocks)

PR target

Branch PRs to feat/mlflow-enterprise-patterns (not main).

Start by reading applications/mlflow/charts/mlflow/values.yaml and applications/mlflow/charts/mlflow/templates/deployment.yaml on the feat/mlflow-enterprise-patterns branch. Add the disabled-by-default GPU settings, conditionally render node selectors, tolerations, and nvidia.com/gpu limits, and include vendor adaptation comments; done means the chart supports the requested GPU pattern without changing default behavior.