Bug: kubeadm join fails with newline-concatenated node names in cluster mode

[ArangoGutierrez]

Description

Post-merge E2E cluster && dedicated test fails because two node names are concatenated with a literal \n instead of being treated as separate entries.

Error

Error from server (NotFound): nodes "ip-10-0-1-8\nip-10-0-1-81" not found
[FAILED] Failed to provision cluster
Expected success, but got an error:
    failed to configure nodes: failed to configure nodes: Process exited with status 1

Root Cause

The node listing/joining logic aggregates node names from SSH output but includes a trailing newline in at least one name, causing two names to be concatenated into a single invalid string when passed to kubectl get node.

Impact

Cluster-mode E2E tests fail intermittently. The bug appears to be in the cluster provisioning code where node names are collected from SSH command output.

Steps to Reproduce

1. Run cluster-mode E2E test (cluster && dedicated label)
2. Observe kubeadm join step fails with concatenated node names

CI Reference

• Workflow run: https://github.com/NVIDIA/holodeck/actions/runs/23744993115
• Job: cluster && dedicated