Improve workaround applied for bug: Multus secondary interface missing on pods after node rolling update

Related to the workaround we applied for Multus secondary interface missing on pods after node rolling update.

We saw the workaround being efficient for node rolling updates and node restarts. The logic of the workaround was to always create a dummy multus-placeholder /etc/cni/net.d/00-multus.conf file on the node, to delay any pod creation until Multus is fully ready.

But I realized that in the current form of the workaround, when Multus unit is installed on the cluster fresh install, we are creating a /etc/cni/net.d/00-multus.conf dummy multus-placeholder and pod creation will fail until Multus is fully installed, leading to a period of time in which pods cannot be created.

So I think we need an extra check in the create-dummy-multus-conf Multus init container, to skip creating the multus-placeholder if Calico .conflist file is already present, something like:

-                                if [ ! -f /host/etc/cni/net.d/00-multus.conf ]; then
+                                if find /host/etc/cni/net.d/ -maxdepth 1 \
+                                        \( -name "*.conf" -o -name "*.conflist" \) \
+                                        ! -name "00-multus.conf*" \
+                                        | grep -q .; then
+                                  echo "Primary CNI config already present, skipping placeholder"
+                                elif [ ! -f /host/etc/cni/net.d/00-multus.conf ]; then