Fix GapFill with ReScan

The GapFill node was not fully reset on a ReScan, so if there was a
GapFill within a NestedLoop, only the first iteration would return
results. This commit fixes this issues.

CHANGELOG.md

@@ -13,6 +13,7 @@ accidentally triggering the load of a previous DB version.**
 **Bugfixes**
 * #1648 Drop chunks for materialized hypertable
 * #1665 Add ignore_invalidation_older_than to timescaledb_information.continuous_aggregates view
+* #1674 Fix time_bucket_gapfill's interaction with GROUP BY
 
 **Thanks**
 * @RJPhillips01 for reporting an issue with drop chunks.

tsl/src/nodes/gapfill/exec.c

@@ -728,6 +728,7 @@ gapfill_rescan(CustomScanState *node)
 	{
 		ExecReScan(linitial(node->custom_ps));
 	}
+	((GapFillState *) node)->state = FETCHED_NONE;
 }
 
 static void

tsl/test/expected/gapfill.out

@@ -2197,6 +2197,27 @@ GROUP BY 3,4;
           5 |    2 | red   |    4
 (10 rows)
 
+-- test with Nested Loop
+SELECT l.id, bucket, data_value FROM
+    (VALUES (1), (2), (3), (4)) a(id)
+    INNER JOIN LATERAL (
+        SELECT b.id id, time_bucket_gapfill('1'::int, time, start=>'1'::int, finish=> '5'::int) bucket, locf(last(data, time)) data_value
+        FROM (VALUES (1, 1, 1), (1, 4, 4), (2, 1, -1), (2, 4, -4)) b(id, time, data)
+        WHERE a.id = b.id
+        GROUP BY b.id, bucket
+    ) as l on (true);
+ id | bucket | data_value 
+----+--------+------------
+  1 |      1 |          1
+  1 |      2 |          1
+  1 |      3 |          1
+  1 |      4 |          4
+  2 |      1 |         -1
+  2 |      2 |         -1
+  2 |      3 |         -1
+  2 |      4 |         -4
+(8 rows)
+
 -- test prepared statement
 PREPARE prep_gapfill AS
 SELECT

tsl/test/sql/gapfill.sql

@@ -1323,6 +1323,16 @@ SELECT
 FROM (VALUES (1,'blue',1),(2,'red',2)) v(time,color,value)
 GROUP BY 3,4;
 
+-- test with Nested Loop
+SELECT l.id, bucket, data_value FROM
+    (VALUES (1), (2), (3), (4)) a(id)
+    INNER JOIN LATERAL (
+        SELECT b.id id, time_bucket_gapfill('1'::int, time, start=>'1'::int, finish=> '5'::int) bucket, locf(last(data, time)) data_value
+        FROM (VALUES (1, 1, 1), (1, 4, 4), (2, 1, -1), (2, 4, -4)) b(id, time, data)
+        WHERE a.id = b.id
+        GROUP BY b.id, bucket
+    ) as l on (true);
+
 -- test prepared statement
 PREPARE prep_gapfill AS
 SELECT