Copy the manager script in two locations to satisfy the operator

We have bumped into an error when the initdb job pod starts:

Error: failed to start container "initdb": Error response from daemon: OCI runtime create failed: container_linux.go:377: starting container process caused: exec: "/controller/manager": stat /controller/manager: no such file or directory: unknown

This is due to the fact that the pod bootstrap-controller initContainer copies the /manager binary into the /controller shared volume, but in our case, this "binary" is only a script running exec /srv/app/manager, which itself is not present in the postgres image (used by the main container).

Init Containers:
  bootstrap-controller:
    Image:         docker-registry.wikimedia.org/repos/data-engineering/postgresql-kubernetes/cloudnative-pg:9b4b94f8748f97c0f3551c72a2b45ed59940ddb8-operator@sha256:3451babc37ff283007b0f0397ba6925d4f44a55e18b62f4dd2352e2e3df690e3
    ...
    Command:
      /manager
      bootstrap
      /controller/manager
      --log-level=info
    ...
    Mounts:
      /controller from scratch-data (rw)
      ...
Containers:
  initdb:
    Image:         docker-registry.wikimedia.org/repos/data-engineering/postgresql-kubernetes/postgresql:15
    ...
    Command:
      /controller/manager
      instance
      init
      ...
    Mounts:
      /controller from scratch-data (rw)

The copying happens here

The /manager command abspath is hardcoded here

Bug: T364797