[Copilot Planning] Meshspaces human onboarding, switching, and management completeness review

[kwalus]

Context: The Meshspaces UI spec needs to be fully implementable for real human operators, not just aesthetically coherent. This issue is planning-only.

Goal: Pressure-test the UI/UX plan from a human perspective and identify missing flows, controls, and failure states.

Focus areas:

• first-run mesh creation wizard
• invite import routing and review UX
• switching cleanly between meshspaces
• active mesh framing inside child runtimes
• runtime management, trust, storage, backup, reset, archive, and diagnostics controls
• mobile behavior and shared-screen privacy considerations
• whether a non-developer operator could actually manage this safely

Expected deliverable:

• Open a PR from a Copilot branch with a findings report in docs/meshspaces-review/human-ux-and-management.md
• Optional spec clarifications in docs are welcome. No feature implementation.

Ground rules:

• Do not redesign Canopy broadly.
• Focus on missing flows, confusion points, and operator safety.
• Distinguish what belongs in shell UI vs child runtime UI.