restructure ux rendering flow and ui bugs for single agent harness

[Douglasymlai]

Related Issue

Description

Adds CAMEL-based single-agent mode and fixes the session/chat UX so single-agent runs behave correctly end-to-end (no workforce-only planning UI, no mode flicker on load).

Backend

• Introduces single_agent_service, factory, toolkits (observable_todo, depth_limited_agent), agent memory, and chat routing so tasks can run as a single CAMEL agent instead of a workforce.

Frontend — session mode

• Adds inferSessionModeFromTask() with a null fallback while a saved session is still loading, so the UI does not flash workforce → single-agent.
• Defaults new projects to Single Agent (pageTabStore v1 migration clears legacy persisted workforce default).
• Hides the session side panel until mode is known on an existing session; workspace/chat input still resolve to a concrete mode for starting tasks.
• Skips pre-setting sessionMode on replay/share playback; mode is inferred from streamed events.

Frontend — single-agent chat flow

• Single-agent tasks skip planning/splitting/confirm phases (UserQueryGroup, taskLifecycleUi, sessionNavLead).
• Shows the work-log area and “Preparing to execute” during PENDING (before first todo_state), matching direct-run behavior.
• Starts the work-log timer on first todo_state (no confirm step).
• Renames display label Single Agent → CAMEL Agent across chat store and types.

Frontend — work log & polish

• TaskWorkLogAccordion: singular “Preparing agent” label, unified running header shimmer, capitalized narration.
• Sidebar icons for single-agent follow task status (no HITL/hand-from-waiting subtasks).
• Unit test for single-agent preparation block label.

Testing Evidence (REQUIRED)

Automated

# Backend
cd backend && pytest tests/app/agent/toolkit/test_observable_todo_toolkit.py tests/app/agent/toolkit/test_depth_limited_agent_toolkit.py tests/app/utils/test_agent_memory.py -q

# Frontend
npm run test -- test/unit/components/TaskWorkLogAccordion.test.tsx

Manual (attach screenshots/screen recording to this PR)

• New task — Single Agent (default): From Workspace, start a task in Single Agent mode → no plan/split card; “Preparing to execute” then work log appear while status is still pending; side panel shows Single Agent view.

• New task — Workforce: Toggle to Workforce → plan/split/confirm flow unchanged.

• Reload saved single-agent session: Open an existing single-agent chat → side panel does not flash workforce UI before settling on Single Agent.

• Reload saved workforce session: Open workforce session → workforce side panel and splitting/HITL icons behave as before.

• Replay/share: Play back a saved session → session mode and side panel match playback events (no wrong-mode flash at start).

• Work log: Running step shows full header shimmer; prep block reads “Preparing agent” in single-agent mode.

• I have included human-verified testing evidence in this PR.

• This PR includes frontend/UI changes, and I attached screenshot(s) or screen recording(s).

• No frontend/UI changes in this PR.

What is the purpose of this pull request?

• Bug fix
• New Feature
• Documentation update
• Other

Contribution Guidelines Acknowledgement

• I have read and agree to the Eigent Contribution Guideline