Document common mecatui delegation workflows

Summary

Document common, real-world mecatui delegation workflows for people running Mecatl locally or connecting mecatui to a shared remote service. This is independent of the public-documentation IA restructuring.

The goal is to help people decide when to use their main session, Subagents, Parallel branches, or Teams; start the right mode through a normal prompt; and monitor the work from mecatui.

Proposed documentation

Add a task-oriented mecatui guide (for example, Work with delegated agents) that:

• gives a concise decision guide for the main session, Subagents, Parallel, and Teams;
• explains each mode through realistic workloads, including bounded investigation, independent implementation, and a read-only panel review of a shared working tree or diff;
• provides clearly labeled Copy this prompt into Mecatl examples, including an optional mode-selection prompt that asks Mecatl to recommend a mode before it changes anything;
• explains that people initiate delegation through a normal prompt rather than a dedicated TUI workflow;
• documents live supervision in mecatui: f6 for the Agents overlay, /team for its Teams tab, footer counts, ctrl+t for bounded child/tool details, and x to cancel a background child;
• distinguishes /agents (named-agent-definition inventory) from the live delegation monitor;
• states that connected-server capabilities and policy govern availability for remote sessions.

Scope and ownership

Keep the guide procedural and user-task-focused under user-docs/mecatui/. Link to the canonical capability material in user-docs/features/subagents-and-teams.md for lifecycle, workspace, isolation, direct-write, merge, and API semantics. Do not duplicate those implementation details per workflow.

Non-goals

• Change delegation behavior or build a new UI.
• Document builder extension points or SDK APIs.
• Treat Parallel temporary workspaces as a persistent manual-review workflow.

Acceptance criteria

• A reader can select an appropriate delegation mode for at least investigation, isolated implementation, and collaborative/panel review work.
• Every documented guided task includes an actionable prompt and a way to monitor or cancel the work in mecatui.
• The guide accurately distinguishes /agents from the f6 live Agents overlay.
• Shared delegation semantics remain canonical in the Capabilities documentation rather than being cloned into the guide.