feat: add error_details to batch evaluation response

- Shows first 10 errors with activity_id, training_type_id, and error message
- Helps debug evaluation failures

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

backend/evaluation_helper.py

@@ -228,6 +228,9 @@ def batch_evaluate_activities(
         "errors": 0
     }
 
+    # Track error details
+    error_details = []
+
     for activity in activities:
         activity_dict = dict(activity)
         try:
@@ -246,7 +249,16 @@ def batch_evaluate_activities(
 
         except Exception as e:
             logger.error(f"[BATCH-EVAL] Error evaluating {activity_dict['id']}: {e}")
+            error_details.append({
+                "activity_id": activity_dict['id'],
+                "training_type_id": activity_dict.get('training_type_id'),
+                "error": str(e)
+            })
             stats["errors"] += 1
 
+    # Add error details to stats (limit to first 10)
+    if error_details:
+        stats["error_details"] = error_details[:10]
+
     logger.info(f"[BATCH-EVAL] Completed: {stats}")
     return stats