Deleting the /var/lib/docker directory will not upgrade Docker Engine CE to Docker Engine EE. It will only remove all the data stored by Docker, such as images, containers, volumes, and networks. This can cause data loss and disrupt the running services. To upgrade from Docker Engine CE to Docker Engine EE, you need to follow the official instructions from Docker, which involve uninstalling the CE package and installing the EE package. You also need to have a valid license to use Docker Engine EE. References:
* Upgrading Docker CE to EE for the Impatient - Part I
* Install Docker Engine
* Moving from docker ce to docker-EE - Stack Overflow

The statement is not correct. A Dockerfile does not store the Docker daemon's configuration options. A Dockerfile is a text document that contains all the commands a user could call on the command line to assemble an image1. A Dockerfile is used to build images, not to configure the Docker daemon. The Docker daemon's configuration options are stored in a JSON file, which is usually located at /etc/docker/daemon.json on Linux systems, or C:\ProgramData\docker\config\daemon.json on Windows2. The JSON file allows you to customize the Docker daemon's behavior, such as enabling debug mode, setting TLS certificates, or changing the data directory2. References: Dockerfile reference), Docker daemon configuration overview)

Explanation
The networkPolicy will block the traffic from a pod bearing the tier: backend label, to a pod bearing the tier:
frontend label. The networkPolicy specifies that only pods with the tier: frontend label can access the pods with the app: guestbook-api and tier: backend labels on port 801. Any other traffic to the backend pods will be denied by default2. Therefore, a request issued from a pod bearing the tier: backend label, to a pod bearing the tier: frontend label will be blocked by the networkPolicy. References: Connect a Frontend to a Backend Using Services), Network Policies)

= This is not how the seven managers should be distributed across three datacenters or availability zones. A swarm cluster is a group of Docker hosts that are running in swarm mode and act as managers or workers1. A manager node is responsible for maintaining the swarm state and orchestrating the services2. A swarm cluster needs a quorum of managers to operate, which means a majority of managers must be available and able to communicate with each other3.
The problem with distributing the seven managers as 4-2-1 is that it creates a split-brain scenario, where the cluster can lose the quorum if one datacenter or availability zone fails. For example, if the datacenter with four managers goes down, the remaining three managers will not have enough votes to form a quorum, and the cluster will stop functioning. Similarly, if the datacenter with one manager goes down, the cluster will lose the tie-breaking vote and will not be able to elect a leader4.
A better way to distribute the seven managers across three datacenters or availability zones is to use 3-2-2, which ensures that the cluster can tolerate the failure of any one datacenter or availability zone and still maintain the quorum. For example, if the datacenter with three managers goes down, the remaining four managers will have enough votes to form a quorum and elect a leader. Similarly, if the datacenter with two managers goes down, the remaining five managers will have enough votes to form a quorum and elect a leader4. References:
* Swarm mode overview | Docker Docs
* Administer and maintain a swarm of Docker Engines | Docker Docs
* Raft consensus in swarm mode | Docker Docs
* Docker Swarm: How to distribute managers across availability zones? - Stack Overflow