Update the logic in the Nomad client's alloc health tracker which erroneously marks existing healthy allocations with dead poststart ephemeral tasks as unhealthy even if they were already successful during a previous deployment.
