Allow termination grace period to be defined per job

[musse]

Is your feature request related to a problem? Please describe.
Infrabox does not let its jobs gracefully terminate, meaning they can not run any needed tear down procedures if the job is terminated prematurely (e.g. job aborted or restarted). The grace period is hard-coded to zero on the coding. The k8s default is 30 seconds. Any clean up must be executed on a separate job, which is error-prone and may require passing data from one job to another.

Describe the solution you'd like
Make the grace period configurable so that an Infrabox job can catch the termination signal when it is being deleted and clean up after itself. Add a field to the job definition to define the grace period and use it when creating the k8s job.