Reporting streaming job metrics

Spark structured streaming provides a simple metric for each small batch, using them it is possible to monitor a streaming job.

https://spark.apache.org/docs/latest/structured-streaming-programming-guide.html#monitoring-streaming-queries

We can simply report metrics using drop wizard library via the metricsEnabled setting, but I think it would be convenient to implement a StreamingQueryListener so that only the necessary events are taken into the callback.

Personally, if we implement a KafkaStreamingListener and send metrics to Kafka, it seems to be easy to save in other storage or configure dashboards or alarms.