Protocol Support for Hot Reload with State Migration

[tercel]

[Problem]
In dynamic AI environments, module logic often needs to be updated without restarting the host process (hot-reloading). However, simply replacing the module instance causes any transient "in-memory" state (like a session buffer or a cache) to be lost.

[Why]
Zero-downtime updates are critical for production-grade AI agents. If an agent is in the middle of a complex multi-step task, restarting the service would break the chain of thought and lose context.

[How]

• Lifecycle Hooks: Added on_suspend() and on_resume() to the Module interface.
• Migration Algorithm: Defined a formal state-transfer protocol where the old instance serializes its state into a StateContainer which the new instance consumes upon loading.
• Renumbering: Updated the Protocol Specification to accommodate the new hot-reload phase in the module lifecycle.