1- Back-to-back invocation of argo_nodeos_config for container manipulation 
   could fail. This is due to the cgroup filesystem (sysfs in general) taking
   some time to do its processing. It is recommended to use sync or put some
   sleep between the invocations.

2- While altering a container, argo_nodeos_config is not particularly forgiving
   of "mindless" user inputs. Here are a few examples:
   	-If one tries to add to a container a resource that it already has, the
	operation will fail (with an explicit error message). The same happens for a
	resource being removed. This should be changed to exhibit an idempotent
	behavior instead (with respect to the particular hardware resource being
	added or removed).
	Basically, if a container has the cpus cores {0,1,4,5}; an "alter_container"
	command that tries to add {0,1,8,9} will fail because of the attempt to
	"re-add" {0,1}. Ideally, the operation should actually succeed and lead the
	container to now possess {0,1,4,5,8,9}.

3- alter_service_os is partially implemented (and is not working yet). Instead,
   the user can delete and recreate (in the meantime).

4- json files as input was planned from the beginning; but is not implemented
   (didn't see this as a high priority feature).