Enhance CI e2e-hub test reliability

This change updates the CI e2e-hub test to wait for the `clusteradm join`
command to succeed asynchronously. Waiting for the latter synchronously
leads some times errors like the following:

```log
+ clusteradm join --hub-token *** --hub-apiserver \
    https://192.168.49.2:8443 --wait --cluster-name minikube
  Error: no matches for kind "Klusterlet" in version "operator.open-cluster-management.io/v1"
```

It also removes the unneeded `curl` installation. It also remove the
`apt-get upgrade` that could lead to more reliability issues on the CI
e2e steps.

open-cluster-management-io/clusteradm#334 (comment)

Signed-off-by: Michail Resvanis <mresvani@redhat.com>

.github/workflows/e2e.yaml

@@ -145,9 +145,6 @@ jobs:
 
     steps:
 
-      - name: Install curl on the machine
-        run: sudo apt update -y && sudo apt upgrade -y && sudo apt install -y curl
-
       - name: Checkout
         uses: actions/checkout@v3
 

ci/prow/e2e-hub-spoke-incluster-build

@@ -24,11 +24,15 @@ hub_api_server=$(kubectl config view -o yaml | yq '.clusters[] | select(.. | .na
 clusteradm join \
   --hub-token ${token} \
   --hub-apiserver ${hub_api_server} \
-  --wait \
   --cluster-name ${MINIKUBE}
+
+# Wait for the join command to succeed
+kubectl wait --for=condition=Established crd managedclusters.cluster.open-cluster-management.io
+timeout 60 sh -c "while ! kubectl get managedcluster ${MINIKUBE} >/dev/null 2>&1; do sleep 2; done"
+timeout 30 sh -c "while ! kubectl get csr -l open-cluster-management.io/cluster-name=${MINIKUBE} 2> /dev/null | grep Pending >/dev/null 2>&1; do sleep 2; done"
 kubectl wait --for=condition=Available deployment/klusterlet-registration-agent -n open-cluster-management-agent
-clusteradm accept --clusters ${MINIKUBE}
 
+clusteradm accept --clusters ${MINIKUBE}
 
 # Installing required addons to the hub/spoke clusters
 clusteradm install hub-addon --names application-manager