feat: progressive multi-round face matching + pending person API

- Identity agent: per-face max matching, multi-round with derived
  seeds from high-confidence faces, angle diversity filter (cosine sim < 0.90)
- Pending person API: POST /file/:file_uuid/pending-person
  + GET /file/:file_uuid/pending-persons with status=pending, source=manual
- Update API docs (07_identity.md)

scripts/fix_processor_stats.py

@@ -0,0 +1,74 @@
+#!/usr/bin/env python3
+"""
+Fix processor_results statistics (frames_processed, output_size_bytes)
+for all jobs that have missing data due to Worker crash.
+"""
+
+import os
+import sys
+import re
+import psycopg2
+
+def fix_processor_stats():
+    conn = psycopg2.connect(os.environ.get('DATABASE_URL', 'postgres://accusys@localhost:5432/momentry'))
+    cur = conn.cursor()
+    
+    # Find jobs with missing frames_processed
+    cur.execute("""
+        SELECT DISTINCT mj.uuid, mj.id
+        FROM public.monitor_jobs mj
+        JOIN public.processor_results pr ON pr.job_id = mj.id
+        WHERE pr.frames_processed = 0
+        AND pr.status = 'completed'
+        ORDER BY mj.created_at DESC
+    """)
+    
+    jobs = cur.fetchall()
+    print(f"Found {len(jobs)} jobs with missing statistics")
+    
+    for uuid, job_id in jobs:
+        print(f"\nProcessing UUID: {uuid}")
+        
+        # Get total_frames from YOLO output
+        yolo_file = f'/Users/accusys/momentry/output/{uuid}.yolo.json'
+        total_frames = 0
+        
+        if os.path.exists(yolo_file):
+            with open(yolo_file, 'r') as f:
+                content = f.read(5000)
+                match = re.search(r'"total_frames": (\d+)', content)
+                if match:
+                    total_frames = int(match.group(1))
+                    print(f"  total_frames: {total_frames}")
+        
+        if total_frames > 0:
+            # Update frames_processed
+            cur.execute("""
+                UPDATE public.processor_results 
+                SET frames_processed = %s
+                WHERE job_id = %s
+            """, (total_frames, job_id))
+            
+            # Update output_size_bytes for each processor
+            processors = ['asr', 'yolo', 'face', 'ocr', 'pose', 'cut', 'appearance', 'asrx']
+            for proc in processors:
+                file_path = f'/Users/accusys/momentry/output/{uuid}.{proc}.json'
+                if os.path.exists(file_path):
+                    size = os.path.getsize(file_path)
+                    cur.execute("""
+                        UPDATE public.processor_results 
+                        SET output_size_bytes = %s
+                        WHERE job_id = %s AND processor_type = %s
+                    """, (size, job_id, proc))
+                    print(f"  {proc}: {size} bytes")
+            
+            conn.commit()
+            print(f"  ✓ Updated")
+        else:
+            print(f"  ⚠ Skipped (no total_frames)")
+    
+    conn.close()
+    print(f"\nCompleted: {len(jobs)} jobs processed")
+
+if __name__ == '__main__':
+    fix_processor_stats()