Make use of Jobs to manage chaincode workload

[jt-nti]

The k8s builder currently creates pods to run chaincode directly, which leads to some problems managing and clearing up pods because the Fabric builder and launcher framework doesn't quite match up with the usual kubernetes workload management. See issues #88, #98 and #100.

A possible improvement would be to make use of kubernetes jobs to run chaincode instead. Using Automatic Cleanup for Finished Jobs and always starting a new job in the run binary should simplify the current implementation and reduce the amount of manual cleanup required in kubernetes.