Prometheus metrics. Configurable timeout and memory limit

[andreygolev]

Here were have:

• configurable timeout for job execution
• configurable memory limit for a shell job
• prometheus realtime cpu and memory consumption while it's working
• prometheus push style metrics after job execution. Useful for monitoring if job was executed in time (that's how we found misses of executions). Also there: exit code, duration of a job, last execution timestamp,

In cases anyone needs it, we can add a flag to turn on/off metrics expose through prometheus.

[vcastellm]

This is a great contribution thanks! But can you split this in several PRs each one with one of the features?