use row_number to dedupe (#798)

tests/test_silver__transactions_and_votes_missing_7_days.sql

@@ -21,6 +21,10 @@ silver_counts AS (
         count(tx_id) AS transaction_count
     FROM
         (
+            -- Deduplicate using row_number() instead of `UNION` or `count(DISTINCT)` since it is faster for large datasets
+            SELECT 
+                *
+            FROM (
                 SELECT
                     block_id,
                     tx_id
@@ -28,7 +32,7 @@ silver_counts AS (
                     {{ ref('silver__transactions') }} t
                 WHERE
                     block_timestamp BETWEEN current_date - 8 AND current_timestamp - INTERVAL '12 HOUR'
-            UNION
+                UNION ALL
                 SELECT
                     block_id,
                     tx_id
@@ -37,6 +41,9 @@ silver_counts AS (
                 WHERE
                     block_timestamp BETWEEN current_date - 8 AND current_timestamp - INTERVAL '12 HOUR'
             )
+            QUALIFY
+                row_number() OVER (PARTITION BY tx_id ORDER BY block_id) = 1
+        )
     GROUP BY
         1
 )