Restore archived tasks, fix dashboard scroll-reset, bind ornith, add Playwright smoke test

- **Restore 82 completed tasks** from tasks/complete/ back to tasks/ top
  level (all <7 days old per the cleanup policy; premature bulk archive
  was fixed).
- **Dashboard: fix scroll-reset on auto-refresh** — renderBoard rebuilds
  the board via innerHTML every 2s, destroying each column-body's
  scrollTop. Now snapshots column-body scrollTop + board.scrollLeft +
  view.scrollTop before rebuild and restores after (matched by
  PHASE_GROUPS index).
- **Dashboard UI additions** (pre-existing unstaged work): approval
  section cards, transition buttons, inline artifact editor (textarea for
  writing missing SPEC/VERDICT/etc from the detail modal).
- **Bind ornith as Implement model** — config.md: Model explicit to
  omlx/Ornith-1.0-35B-4bit-mlx, context window 32768. Interactive
  autopilot already used ornith via opencode default; now explicit.
- **Fix cleanup stub** — automaton-cleanup.sh had a stale --project arg
  pointing at a pytest temp dir (test isolation leak). Rewired to point
  at ~/.automaton.
- **Fix plist-isolation test** — test asserted host plist doesn't exist,
  but a real install creates it. Now snapshots mtime before run, asserts
  unchanged after (only a write during the test counts as bleed).
- **New Playwright smoke test** (tests/test_dashboard_ui.py) — 2 tests:
  board renders tasks, column scroll survives auto-refresh tick.
  Verified the test fails without the scroll fix (scrollTop resets to 0).
  Skipped via importorskip when playwright is absent (main CI stays
  green).
- **Clarify SI loop scope in README** — new-project onboarding section
  documents the framework-scoped self-improvement loop and options
  (leave/pause/create project loop).
- **CHANGELOG** documents all changes including the known model-divergence
  gap (mde tasks marked complete but per-role model binding was never
  implemented).

tasks/task-status-reason/SPEC.md

@@ -0,0 +1,40 @@
+# Add Status Reason to Task Display
+
+## Goal
+Show a human-readable explanation of *why* a task is in its current state, fix the pending review count to exclude terminal-state tasks, and replace approve/request-changes buttons with revoke buttons when a review is already submitted.
+
+## Requirements
+
+### R1. Add `status_reason` property to Task
+Add a `status_reason` property on the `Task` dataclass that derives a human-readable explanation from the task's state, artifacts, and verdict status. For example:
+- DONE → "Verdict: PASS"
+- BLOCKED → "Verdict: FAIL — changes required before re-review"
+- BUG_FIND (with IMPLEMENTATION.md) → "Implementation complete — awaiting bug finding"
+- BACKLOG → "No artifacts yet — not started"
+
+### R2. Expose `status_reason` in API responses
+Include `status_reason` in the JSON response for both `/api/tasks` and `/api/task/{name}`.
+
+### R3. Display status reason prominently
+- Show `status_reason` below the status badge in the task detail panel
+- Show `status_reason` on task cards for blocked/bug_find/adv_bug_find states
+
+### R4. Fix pending review count
+The pending review counter in the header currently counts all tasks without a review, including DONE tasks. Fix it to exclude terminal states (done, blocked) so only active tasks count toward pending.
+
+### R5. Replace review buttons with revoke on submitted reviews
+When a task already has an approved review, replace the approve button with "Revoke Approval". When changes are requested, replace with "Revoke Changes". Both send status="pending" to reset the review.
+
+## Acceptance Criteria
+- [ ] `status_reason` property works for all task states
+- [ ] `/api/tasks` responses include `status_reason`
+- [ ] Detail panel shows status reason below the status badge
+- [ ] Task cards show status reason for blocked/bug_find/adv_bug_find tasks
+- [ ] Pending review count excludes done/blocked tasks
+- [ ] Approved tasks show "Revoke Approval" instead of "Approve"
+- [ ] Backend accepts "pending" as a review status
+- [ ] All tests pass
+
+## Non-Goals
+- Not changing the task state machine — revoking a review only resets the review metadata, not the task state
+- Not adding buttons to change task state directly (e.g., send back to planning) — that's a larger feature