Don't rely on pg_stat_wal_receiver when deciding on pg_rewind (#2863)

As was reported by @ants on Slack it could happen that `received_tli` is ahead of replayed timeline, therefore we should stop using it when deciding on pg_rewind if postgres is running and use only `IDENTIFY_SYSTEM` via replication connection.

patroni/postgresql/rewind.py

@@ -158,7 +158,7 @@ class Rewind(object):
     def _get_local_timeline_lsn(self) -> Tuple[Optional[bool], Optional[int], Optional[int]]:
         if self._postgresql.is_running():  # if postgres is running - get timeline from replication connection
             in_recovery = True
-            timeline = self._postgresql.received_timeline() or self._postgresql.get_replica_timeline()
+            timeline = self._postgresql.get_replica_timeline()
             lsn = self._postgresql.replayed_location()
         else:  # otherwise analyze pg_controldata output
             in_recovery, timeline, lsn = self._get_local_timeline_lsn_from_controldata()

tests/test_rewind.py

@@ -93,7 +93,7 @@ class TestRewind(BaseTestPostgresql):
 
         with patch.object(Postgresql, 'is_running', Mock(return_value=True)), \
                 patch.object(MockCursor, 'fetchone',
-                             Mock(side_effect=[(0, 0, 1, 1, 0, 0, 0, 0, 0, None, None, None), Exception])):
+                             Mock(side_effect=[Exception, (0, 0, 1, 1, 0, 0, 0, 0, 0, None, None, None)])):
             self.r.rewind_or_reinitialize_needed_and_possible(self.leader)
 
     @patch.object(CancellableSubprocess, 'call', mock_cancellable_call)