ignore not succeeded executions in cost calculation (#36)

We don't want to include broken executions in the node cost calculation; not succeeded node executions normally don't have an realistic execution time.

Author: Leonhard Schick

data_integration/logging/node_cost.py

@@ -32,6 +32,7 @@ def node_durations_and_run_times(node: pipelines.Node) -> {tuple: [float, float]
         FROM data_integration_node_run node
         WHERE node.node_path [ 0 : {'%(level)s'}] = %(path)s
           AND array_length(node.node_path, 1) > {'%(level)s'}
+          AND succeeded = TRUE
         GROUP BY node.node_path)
 SELECT node_path,
        round(avg(avg_duration)::NUMERIC, 1) AS avg_duration,