Workboard v2 — outcome buckets

About this view (Track C, 2026-05-25 plan): Cards are grouped by OUTCOME they drive, not by category. Five buckets: Convert (lead → booking), Recover (ghosting risk → reply), Retain (member at risk → check-in), Close (explicit no → close with reason) — all AI-can-act. Appointment outcome is human-only (attended/no-show — AI can't observe in-person). System cards (lead-capture failures, handler-idle) sit in a separate bucket below. Compare to /tasks; if this reads cleaner, we'll promote.