Extend the stream.sh to track consumer latencies

The idea is to have a command to track periodically the latencies of consumers.
This latencies information can be replicated in a different datacenter and used to position the consumers in case of failover.

The latencies information are send into a Log, (for instance it can be named _consumer_latencies)

The latency is expressed

• with a key composed of:
  • the consumer group
  • log name
  • partition
• and a value composed of:
  • the date where the latency has been measured
  • the timestamp of the last processed record
  • eventually the key of the last processed record

This way another command can use this log to restore the consumer positions at a given time.

The rational for the composed key is to use the Kafka log compaction feature so the _consumer_latencies log can stay compact over time and the command to restore the position can quickly get the latest latency.