Document jobs behavior after failure/timeout

[itrofimow]

Basically what title says.

For now it's unclear what happens if a job fails - would it be rescheduled?
If so, to what time? How many times would it be rescheduled, if ever?
How does a recurrent job behave under same circumstances?
How does a sequence of jobs behave under same circumstances?

What happens if job's execution takes a lot of time? There is something about Obsolete jobs in configuration, what does it mean exactly?

[Bobreshovr]

Hi! first block questions is in this issue #25