Consider adding support to choose dynamic behavior of a service

[totto]

i.e. today's semantics will red-flag nodes which are dead based upon (among others) the local ip-address. For some typical container-based deployments like ECS/kubernetes++ a new update/patch/deployment is a full new container, with new ip. If we add a way to mark which service expect what kind of update/patch behavior - those container-based services will avoid getting a lot of red/dead nodes in Visuale.