Automatically fetch config map values

[esposem]

This PR gets rid of some ConfigMap parameters that must be provided to CAA in order to make it run on Azure/AWS.

More specifically:

• In Azure, ConfigMap was

    cat <<EOF | oc apply -f -
apiVersion: v1
kind: ConfigMap
metadata:
  name: ${CM}
  namespace: ${OP_NAME}
data:
  CLOUD_PROVIDER: "azure"
  VXLAN_PORT: "9000"
  AZURE_INSTANCE_SIZE: "Standard_D8as_v5" #set
  PROXY_TIMEOUT: "15m"
  AZURE_IMAGE_ID: ${AZURE_IMAGE_ID}
  AZURE_SUBNET_ID: ${AZURE_SUBNET_ID}
  AZURE_NSG_ID: ${AZURE_NSG_ID}
  DISABLECVM: "true"
EOF

and becomes now

    cat <<EOF | oc apply -f -
apiVersion: v1
kind: ConfigMap
metadata:
  name: ${CM}
  namespace: ${OP_NAME}
data:
  CLOUD_PROVIDER: "azure"
  VXLAN_PORT: "9000"
  AZURE_INSTANCE_SIZE: "Standard_D8as_v5" #set
  PROXY_TIMEOUT: "15m"
  AZURE_IMAGE_ID: ${AZURE_IMAGE_ID}
  DISABLECVM: "true"
EOF

Current limitations of this patch

azure:

• tested on Azure

Fixes: #1200

[esposem]

@snir911 @bpradipt the solution to the issues mentioned in the PR description are:

• Assume the user provided a AZURE_USER_RESOURCE_GROUP field in the peerpodssecret, which corresponds to the user-created azure RG. This is valid only when using ARO/AKS.
• Query for subnets/vnets/nsg but if more than one result is returned, then return an error and force the user to insert it via the ConfigMap
• This implies (and should be documented somewhere) that for ARO the subnet ID must be provided by the user, as workers and masters are in two different user-created subnets, and we need the worker subnet.

[snir911]

Thanks! i added some comments, also please look at the build and lint actions

[esposem]

Updated to work only on Azure. Managed services like ARO and AKS are left as TODO, since they have different resource groups and the Azure SDK API might not fully support them yet.

[bpradipt]

@esposem can you please rebase this PR ?

[esposem]

@bpradipt I rebased, but it looks like we need more work for this? I understood that this change was not needed if we use IMDS... As @snir911 suggested, maybe I need to close this PR?