Allow locking updated tuples in tuple_update() and tuple_delete()

Currently, in read committed transaction isolation mode (default), we have the
following sequence of actions when tuple_update()/tuple_delete() finds
the tuple updated by concurrent transaction.

1. Attempt to update/delete tuple with tuple_update()/tuple_delete(), which
   returns TM_Updated.
2. Lock tuple with tuple_lock().
3. Re-evaluate plan qual (recheck if we still need to update/delete and
   calculate the new tuple for update).
4. Second attempt to update/delete tuple with tuple_update()/tuple_delete().
   This attempt should be successful, since the tuple was previously locked.

This patch eliminates step 2 by taking the lock during first
tuple_update()/tuple_delete() call.  Heap table access method saves some
efforts by checking the updated tuple once instead of twice.  Future
undo-based table access methods, which will start from the latest row version,
can immediately place a lock there.

The code in nodeModifyTable.c is simplified by removing the nested switch/case.

Discussion: https://postgr.es/m/CAPpHfdua-YFw3XTprfutzGp28xXLigFtzNbuFY8yPhqeq6X5kg%40mail.gmail.com
Reviewed-by: Aleksander Alekseev, Pavel Borisov, Vignesh C, Mason Sharp
Reviewed-by: Andres Freund, Chris Travers

src/include/access/tableam.h

@@ -530,7 +530,8 @@ typedef struct TableAmRoutine
 								 Snapshot crosscheck,
 								 bool wait,
 								 TM_FailureData *tmfd,
-								 bool changingPart);
+								 bool changingPart,
+								 LazyTupleTableSlot *lockedSlot);
 
 	/* see table_tuple_update() for reference about parameters */
 	TM_Result	(*tuple_update) (Relation rel,
@@ -542,7 +543,8 @@ typedef struct TableAmRoutine
 								 bool wait,
 								 TM_FailureData *tmfd,
 								 LockTupleMode *lockmode,
-								 TU_UpdateIndexes *update_indexes);
+								 TU_UpdateIndexes *update_indexes,
+								 LazyTupleTableSlot *lockedSlot);
 
 	/* see table_tuple_lock() for reference about parameters */
 	TM_Result	(*tuple_lock) (Relation rel,
@@ -1457,7 +1459,7 @@ table_multi_insert(Relation rel, TupleTableSlot **slots, int nslots,
 }
 
 /*
- * Delete a tuple.
+ * Delete a tuple (or lock last tuple version if lockedSlot is given).
  *
  * NB: do not call this directly unless prepared to deal with
  * concurrent-update conditions.  Use simple_table_tuple_delete instead.
@@ -1473,6 +1475,8 @@ table_multi_insert(Relation rel, TupleTableSlot **slots, int nslots,
  *	tmfd - filled in failure cases (see below)
  *	changingPart - true iff the tuple is being moved to another partition
  *		table due to an update of the partition key. Otherwise, false.
+ *	lockedSlot - lazy slot to save the locked tuple if should lock the last
+ *		row version during the concurrent update. NULL if not needed.
  *
  * Normal, successful return value is TM_Ok, which means we did actually
  * delete it.  Failure return codes are TM_SelfModified, TM_Updated, and
@@ -1485,15 +1489,17 @@ table_multi_insert(Relation rel, TupleTableSlot **slots, int nslots,
 static inline TM_Result
 table_tuple_delete(Relation rel, ItemPointer tid, CommandId cid,
 				   Snapshot snapshot, Snapshot crosscheck, bool wait,
-				   TM_FailureData *tmfd, bool changingPart)
+				   TM_FailureData *tmfd, bool changingPart,
+				   LazyTupleTableSlot *lockedSlot)
 {
 	return rel->rd_tableam->tuple_delete(rel, tid, cid,
 										 snapshot, crosscheck,
-										 wait, tmfd, changingPart);
+										 wait, tmfd, changingPart,
+										 lockedSlot);
 }
 
 /*
- * Update a tuple.
+ * Update a tuple (or lock last tuple version if lockedSlot is given).
  *
  * NB: do not call this directly unless you are prepared to deal with
  * concurrent-update conditions.  Use simple_table_tuple_update instead.
@@ -1511,7 +1517,9 @@ table_tuple_delete(Relation rel, ItemPointer tid, CommandId cid,
  *	lockmode - filled with lock mode acquired on tuple
  *  update_indexes - in success cases this is set to true if new index entries
  *		are required for this tuple
- *
+ * 	lockedSlot - lazy slot to save the locked tuple if should lock the last
+ *		row version during the concurrent update. NULL if not needed.
+
  * Normal, successful return value is TM_Ok, which means we did actually
  * update it.  Failure return codes are TM_SelfModified, TM_Updated, and
  * TM_BeingModified (the last only possible if wait == false).
@@ -1530,12 +1538,14 @@ static inline TM_Result
 table_tuple_update(Relation rel, ItemPointer otid, TupleTableSlot *slot,
 				   CommandId cid, Snapshot snapshot, Snapshot crosscheck,
 				   bool wait, TM_FailureData *tmfd, LockTupleMode *lockmode,
-				   TU_UpdateIndexes *update_indexes)
+				   TU_UpdateIndexes *update_indexes,
+				   LazyTupleTableSlot *lockedSlot)
 {
 	return rel->rd_tableam->tuple_update(rel, otid, slot,
 										 cid, snapshot, crosscheck,
 										 wait, tmfd,
-										 lockmode, update_indexes);
+										 lockmode, update_indexes,
+										 lockedSlot);
 }
 
 /*