fix(mongodb): reorganize document size filter logic in aggregation

Reorganize the logic for adding a document size filter in the MongoDB aggregation process. The changed aggregation order improves MongoDB scanning performance.

metadata-ingestion/src/datahub/ingestion/source/mongodb.py

@@ -218,21 +218,22 @@ def construct_schema_pymongo(
     """
 
     aggregations: List[Dict] = []
-    if should_add_document_size_filter:
-        doc_size_field = "temporary_doc_size_field"
-        # create a temporary field to store the size of the document. filter on it and then remove it.
-        aggregations = [
-            {"$addFields": {doc_size_field: {"$bsonSize": "$$ROOT"}}},
-            {"$match": {doc_size_field: {"$lt": max_document_size}}},
-            {"$project": {doc_size_field: 0}},
-        ]
 
     if sample_size:
         if use_random_sampling:
             aggregations.append({"$sample": {"size": sample_size}})
         else:
             aggregations.append({"$limit": sample_size})
 
+    if should_add_document_size_filter:
+        doc_size_field = "temporary_doc_size_field"
+        # create a temporary field to store the size of the document. filter on it and then remove it.
+        aggregations.extend([
+            {"$addFields": {doc_size_field: {"$bsonSize": "$$ROOT"}}},
+            {"$match": {doc_size_field: {"$lt": max_document_size}}},
+            {"$project": {doc_size_field: 0}},
+        ])
+
     documents = collection.aggregate(aggregations, allowDiskUse=True)
 
     return construct_schema(list(documents), delimiter)