taint removal doesn't guarantee node has picked up address

[huwcbjones]

We have a workload that requires knowing the node's static IP.
This is fetched from the KubeIP at application startup.

However, we've observed that even when the taint has been removed, it can take a few minutes for the node's status to reflect the IP assignment.
This has led to our app binding to the old IP and causing connectivity issues later on.

Perhaps it was my misinterpretation of the Node Taints section in the readme, but I expected that when the taint was removed everything would be a-okay.

It is 💯 possible to remove the taint after the node's status reports the assigned address as I have hacked together something to test/confirm my theory above this afternoon.
It's not pretty and it's definitely lacking some tests, but I'll open a draft PR and link it to this issue after I click submit.

[huwcbjones]

Draft PR with working GCP implementation available in #158.
Logs from GKE:

2024-10-24 15:16:06.957 kubeip | kubeip agent started
2024-10-24 15:16:07.858 kubeip | deleting public IP address from instance
2024-10-24 15:16:38.039 kubeip | failed waiting for operation operation-$UUID Error="failed to get operation operation-$UUID: Post "https://compute.googleapis.com/compute/v1/projects/$project/zones/$zone/operations/operation-$UUID/wait?alt=json&prettyPrint=false": read tcp 10.x.y.z:52280->108.x.y.z:443: read: connection reset by peer
2024-10-24 15:16:38.253 kubeip | adding public IP address to instance
2024-10-24 15:16:41.926 kubeip | Node is not yet reporting the assigned address
2024-10-24 15:16:41.926 kubeip | retrying after 1m0s
2024-10-24 15:17:41.931 kubeip | Node is not yet reporting the assigned address
2024-10-24 15:17:41.931 kubeip | retrying after 1m0s
2024-10-24 15:18:41.927 kubeip | Node is reporting assigned address
2024-10-24 15:18:41.957 kubeip | taint key removed successfully