Post-script agent orchestration with trigger gating, flow UI, and coverage

[huhuanming]

Summary

• add post-script agent orchestration across CLI and App, including streamed output, steer/interrupt support, and persisted agent_runs
• add task/workspace memory pipeline and profile-backed task identity for agent runs
• introduce pre-script trigger gating (always / preScriptTrue) with shared evaluator behavior in CLI and App
• improve Script Detail UX with a 3-step Agent Flow Gate and editable Agent Trigger selector in GUI
• add conditional branch visualization for preScriptTrue mode (true -> run agent, false -> skip agent)
• document pre-script gate contract in skills/scriptoria/SKILL.md
• expand CI/test coverage for core + CLI behavior around agent execution and trigger flow

Why

These changes make post-script automation deterministic and observable:

• deterministic gate behavior before agent execution
• clear GUI visibility into which step is blocking or branching
• stronger reliability via automated coverage for agent lifecycle and trigger semantics

Validation

• swift build
• swift test (35 tests passed)