feat(commentary): introduce universal rafters-critic for scenario commentary and cadence control

Implement local in-process inference backend for transformers models

feat: require live infrastructure for event store and inference, removing offline mode

feat(observability): instrument models, conductor, memory, ledger (Units 2-6)

feat: add Hugging Face inference backend next to Modal

feat(core): add resilient loop and shared blackboard

feat: Enhance model routing and deployment flexibility

feat: unify model catalogue and self-hosted routing

feat: LiteLLM gateway routing profiles to Modal-served small models

Refactor code structure for improved readability and maintainability