docs(#945): correct actor-model postcard — result-reporting + taxonomy/citation accuracy

[AndriiPasternak31]

Summary

Closes #945 — and finally links it to a PR.

The #945 deliverable (docs/planning/ACTOR_MODEL_POSTCARD.md) already shipped to dev in commit bc7e5fd2, but it rode in via PR #1293 (the #946 pull-pilot PR), which linked Closes #946 and never #945. So #945 stayed open. A line-by-line audit against the issue ACs + referenced design docs + the live codebase confirms the postcard satisfies every acceptance criterion (envelope fields, per-kind payloads, typed reply terminal-reason, result-reporting reframe, both-way cross-refs, BacklogService backlog_metadata compatibility) — but surfaced one factual error and several citation/count nits, fixed here.

Changes (docs-only, 3 files)

ACTOR_MODEL_POSTCARD.md

1. Factual fix — schedule_execution_completed was described as "never emitted." It is emitted by the scheduler (src/scheduler/service.py, several _publish_event sites) for schedule-triggered runs; it's simply not consumed by any agent-facing/MCP/frontend path and not emitted on the agent→agent MCP path the experiment: pull-coordination pilot — route MCP chat_with_agent through the agent queue #946 pilot uses. The polling conclusion is unchanged — only the supporting claim is corrected.
2. Enum mapping — pin the shipped TaskExecutionErrorCode ({TIMEOUT, CAPACITY, AUTH, BILLING, AGENT_ERROR, NETWORK, CIRCUIT_OPEN, RECONCILED, LEASE_EXPIRED}) and make explicit that the contract adds OOM/MAX_TURNS on top, so the divergence is intentional rather than ambiguous.
3. Citation fix — session_id unifies demotion-map security: implement safe tar extraction with symlink/hardlink validation #8–fix: Add ARM64 Alpine compatibility for frontend Docker build #12 (was security: implement safe tar extraction with symlink/hardlink validation #8–feat: SMARTS trading pipeline with Telegram notifications and Miro visualization #13); feat: SMARTS trading pipeline with Telegram notifications and Miro visualization #13 inject_result is causation_id (already stated above); dropped the non-existent field name continue_session.

ACTOR_MODEL_TASK_DEMOTION_MAP.md + TARGET_ARCHITECTURE.md
4. Stale field count — ParallelTaskRequest has 14 fields today (src/backend/models.py:97-112), not 15. The demotion map's own code block already listed 14. Fixed the prose counts ("13 of its 14 optional", "13 silent siblings"), the stale line-ref (:84-99 → :97-112), and the TARGET_ARCHITECTURE.md mirror of the claim.

Verification

• git diff touches only the three planning docs.
• grep -rn schedule_execution_completed src/ confirms the scheduler emitters; TaskExecutionErrorCode members match the added mapping line; ParallelTaskRequest field count verified at src/backend/models.py:97-112 (14).
• Docs-only — no code, no migrations, no Dockerfile surface; no test/verify-local run required.

🤖 Generated with Claude Code

[github-actions]

⚠️ Nightly unit-suite check skipped — merge conflict against dev.

Resolve by running git merge dev locally and pushing the result. The next nightly run will re-test once the conflict is gone.

[vybe]

✅ /validate-pr — APPROVE

Pure docs correction (3 files, docs/planning/*). Closes #945 (type-docs, same-repo → auto-promotes to status-in-dev on merge). No code, no security/packaging surface. Corrects postcard result-reporting + stale ParallelTaskRequest field count (15→14). All 4 required checks green.