MCP: Disable shutting down Hyperd after a period of disuse

Hyperd doesn't take much resources when not in use and when the MCP daemon shuts it down and then needs to restart it, the user gets an error and a message saying to retry their tool usage. Since there is little cost in keeping hyper running, let's just remove the auto shutdown.

Resolved by #115 — the daemon is now resident by default (idle timeout disabled unless explicitly set via --idle-timeout or HYPERDB_DAEMON_IDLE_TIMEOUT).