Fix stale read on CreateTranscriptions

arkindex/documents/serializers/ml.py

@@ -416,10 +416,16 @@ class TranscriptionBulkSerializer(serializers.Serializer):
 
     def validate(self, data):
         element_ids = set(transcription['element_id'] for transcription in data['transcriptions'])
-        found_ids = set(Element.objects.filter(
-            id__in=element_ids,
-            corpus__in=Corpus.objects.writable(self.context['request'].user)
-        ).values_list('id', flat=True))
+        found_ids = set(
+            Element
+            .objects
+            .using('default')
+            .filter(
+                id__in=element_ids,
+                corpus__in=Corpus.objects.writable(self.context['request'].user)
+            )
+            .values_list('id', flat=True)
+        )
 
         missing_ids = element_ids - found_ids
         if not missing_ids: