Support disabling or working with core dumps

[dhrp]

On most? Kubernetes distributions core dumps are enabled. But if a process exists the dumpfile can easily be in the GB's. When a pod runs in a space-restrained environment it is possible that this fills up the available disk space and subsequent tasks that moulin runs fail, or the pod is killed because of out of ephemeral storage issues.

Ideas for improvement:

• Allow disabling core dumps altogether ulimit -c 0
• Delete core dumps. In /proc/sys/kernel/core_pattern we can find the pattern under which they are stored
• Allow configuration of where to upload them for introspection.