Fix/review remediation 018 026

[skibkitty]

Summary

Implements TASK-018 through TASK-026 — the full-branch code-review remediation backlog recorded in docs/tasks.md. All nine tasks address review findings across domain immutability, application-layer deduplication, UI maintainability, persistence hardening, the benchmark-report contract, simulator deltas, and documentation accuracy. No behavior changes beyond the documented fixes.

Changes

TASK-018 — Fix recommendation deep-freeze gap

• Deep-frozen the affectedTaskIds arrays on the tie-break-applied and blocked-status-eligible warnings, honoring the ADR-008 immutability contract
• Regression test asserts Object.isFrozen on the nested arrays and elements

TASK-019 — Extract shared task-to-input mapper

• Removed the copy-pasted Task → CreateTaskInput mapper from task-service.ts, goal-service.ts, and dependency-service.ts
• Single shared implementation in src/application/task-input.ts (including the goalId: null → undefined bridge)

TASK-020 — Extract shared test helpers

• Consolidated ~10 duplicated createStubRepository implementations (plus createInMemoryStorage) into src/test-support/index.ts
• All call sites migrated; the shared stub seeds via initialProject and supports per-method overrides
• Net −275 lines of duplicated helper code

TASK-021 — UI status options from domain constant

• TaskForm/TaskList dropdowns now render from the domain's ALL_TASK_STATUSES
• STATUS_COLORS in the dependency graph is keyed by TaskStatus (no typo-masking fallback)
• as TaskStatus casts replaced with a type-safe find()

TASK-022 — Persistence validation and freeze hardening

• Serialization copies and deep-freezes dependencies so serialized output does not alias mutable input
• Deserialization rejects non-string dependency entries with a descriptive error instead of silently filtering
• ADR-010 updated to document the chosen semantics

TASK-023 — Guard against project id collisions

• ProjectService.createProject rejects an existing id (Project already exists: {id}), covering the sample-project seed path
• Dashboard surfaces the error instead of silently overwriting

TASK-024 — Benchmark report emission as a hard contract

• npm run benchmark now fails if it cannot write benchmark/results.txt, matching the CI if-no-files-found: error artifact contract
• Covered by a write-failure simulation test

TASK-025 — Simulator blocked-task and risk deltas

• SimulationResult now exposes blockedTaskCount per side, blockedTaskDelta, and newlyCriticalTaskIds (slack exhausted → becomes critical) as the deterministic risk indicator
• Definitions recorded in ADR-009; the scenario comparison panel displays both
• Backward compatible (new fields only)

TASK-026 — Documentation cleanup

• Archived docs/task-006-review.md (superseded banner, no live references)
• Corrected stale counts (TASK-015: 286→287; handoff engine tests 31→34)
• Added isReachable(t, t) self-reachability note to the case study
• Narrowed ADR-006's complexity claim; pointed TASK-014's determinism criterion at ADR-011
• Removed unused .next/ entry from .gitignore

Additional (not in the task list)

• Integration test locking the demo-story outcome: sample project's deterministic recommendation, factor breakdown, task count, and critical path

Verification

• npm run verify passes (typecheck, 304 tests, lint, format:check)
• npm run build succeeds
• npm run benchmark passes (7 tests, benchmark/results.txt written)
• npm run test:e2e passes (2 Playwright specs)

Docs

• docs/tasks.md — TASK-018 through TASK-026 marked DONE
• docs/decisions.md — ADR-009 (blocked-task/risk deltas), ADR-010 (dependency-entry validation)
• docs/handoff.md — current task and verification updated
• docs/progress.md — remediation-completed entry appended