Add better management of SIGTERM for the elasticsearch child process of run.sh

[primeroz]

Use example from pires/kubernetes-elasticsearch-cluster#146 (comment) to better manage the SIGTERM of the elasticsearch process when the container is TERMed

• trap SIGTERM And term the java process pid
• SLEEP for 15 seconds ( Configurable ) after waiting for PID to terminate to try address issues:
  • Re-election takes over 30 seconds when deleting master pod (but fast when killing the process directly) pires/kubernetes-elasticsearch-cluster#231
  • https://gitlab.com/mintel/satoshi/infrastructure/terragrunt-satoshi-cluster-infrastructure/issues/90
• remove pkilling from "stop" scripts
• remove ability to run process as root , this container is only for running as unprivileged user