feat: Activity Cluster Placeholder Registry - Complete Implementation (17 Placeholders)

Implements complete placeholder registry for Activity & Training metrics following
Phase 0c Multi-Layer Architecture pattern.

SCOPE: 17 Activity Placeholders
- Group 1 (3): Legacy Resolver - activity_summary, activity_detail, trainingstyp_verteilung
- Group 2 (7): Basic Metrics - volume, frequency, quality, load, monotony, strain, rest compliance
- Group 3 (7): Advanced Metrics - 5x ability_balance, vo2max_trend, activity_score

IMPLEMENTATION:
- File: backend/placeholder_registrations/activity_metrics.py (~1,100 lines)
- Pattern: Nutrition Part A (common_metadata + evidence-based tagging)
- Evidence: CODE_DERIVED (58%), DRAFT_DERIVED (16%), MIXED (15%), TO_VERIFY (6%), UNRESOLVED (5%)
- Formulas: All documented in known_limitations (Load Model, Monotony, Strain, Ability Balance, Activity Score)

CRITICAL ISSUES IDENTIFIED (NOT FIXED per NO LOGIC CHANGES):
1. quality_label field mismatch (quality_sessions_pct) - TO_VERIFY
2. RPE moderate quality mapping bug (proxy_internal_load_7d) - CODE_DERIVED
3. JSONB dependencies (6 placeholders) - ability_balance_*, rest_day_compliance
4. vo2max_trend_28d questionable category (Recovery vs. Activity) - TO_VERIFY

TESTING:
✓ All 17 placeholders registered successfully
✓ Registry size: 48 (31 pre-existing + 17 new)
✓ Dev backend integration: no errors
✓ Auto-registration on module import: working

ARCHITECTURE ALIGNMENT:
- Phase 0c Multi-Layer: 14/17 aligned (Group 2 + 3)
- Old Resolver Pattern: 3/17 (Group 1 - documented, should be refactored)
- Layer separation: data_layer → resolver → export

FILES:
- NEW: backend/placeholder_registrations/activity_metrics.py
- MODIFIED: backend/placeholder_registrations/__init__.py (added import)
- MODIFIED: CLAUDE.md (placeholder registry rules)

DOCUMENTATION:
- Gap Analysis: .claude/task/rework_0b_placeholder/ACTIVITY_CLUSTER_GAP_ANALYSIS.md
- Code Inspection: .claude/task/rework_0b_placeholder/ACTIVITY_CLUSTER_CODE_INSPECTION.md
- Implementation Report: .claude/task/rework_0b_placeholder/ACTIVITY_CLUSTER_IMPLEMENTATION_REPORT.md

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

CLAUDE.md

@@ -27,6 +27,14 @@
 - Production Deploy nur nach expliziter Freigabe
 - Migration 001-999 Pattern einhalten
 
+**Placeholder/Metrics (VERBINDLICH ab 2026-04-02):**
+- ✅ **ALLE** neuen Placeholder/Metrics MÜSSEN über Registry Framework registriert werden
+- ✅ Registrierung in `backend/placeholder_registrations/`
+- ✅ Evidence-basiertes Tagging für alle Metadatenfelder (CODE_DERIVED, DRAFT_DERIVED, etc.)
+- ✅ Single Source of Truth für Prompt-Injektion, GUI, Export, Validierung
+- 📚 **Verbindliche Dokumentation:** `.claude/docs/technical/PLACEHOLDER_REGISTRY_FRAMEWORK.md`
+- ❌ **VERBOTEN:** Hardcoded Metadaten, duplizierte Definitionen, Placeholder ohne Registry
+
 ## Projekt-Übersicht
 **Mitai Jinkendo** (身体 Jinkendo) – selbst-gehostete PWA für Körper-Tracking mit KI-Auswertung.
 Teil der **Jinkendo**-App-Familie (人拳道). Domains: jinkendo.de / .com / .life

backend/placeholder_registrations/__init__.py

@@ -9,5 +9,6 @@ from . import nutrition_part_a
 from . import nutrition_part_b
 from . import nutrition_part_c
 from . import body_metrics
+from . import activity_metrics
 
-__all__ = ['nutrition_part_a', 'nutrition_part_b', 'nutrition_part_c', 'body_metrics']
+__all__ = ['nutrition_part_a', 'nutrition_part_b', 'nutrition_part_c', 'body_metrics', 'activity_metrics']