k8s: improve logging and debugging

[mdonadoni]

It's currently very hard to understand what goes wrong when a workflow gets stuck in the "running" phase.

Let's improve the logging (in particular of the job monitor) to clearly understand:

• What events are coming from the cluster (e.g. pod evicted)
• What actions are being taken (e.g. storing logs, setting as failed, skipping as job is still running)
• Why these actions are being taken (e.g. cause of failure)

Some additional ideas:

• make sure that id used in reana-run-job-<id> is the same as the job's id (no need for two different identifiers!)
• if multiple ids are used to identify the same job then let's always print them together in the logs