fix: ignore stale SSH host keys for ephemeral homelab VMs

Fresh destroy/recreate cycles change VM host keys, which was breaking bootstrap after rebuilds. Use a disposable known-hosts policy in the controller SSH options so automation does not fail on expected key rotation.

nixos/kubeadm/bootstrap/controller.py

@@ -110,7 +110,9 @@ class Controller:
             "-o",
             "IdentitiesOnly=yes",
             "-o",
-            "StrictHostKeyChecking=accept-new",
+            "StrictHostKeyChecking=no",
+            "-o",
+            "UserKnownHostsFile=/dev/null",
             "-i",
             self.ssh_key,
         ]
@@ -172,14 +174,7 @@ class Controller:
         return last
 
     def prepare_known_hosts(self):
-        ssh_dir = Path.home() / ".ssh"
-        ssh_dir.mkdir(parents=True, exist_ok=True)
-        (ssh_dir / "known_hosts").touch()
-        run_local(["chmod", "700", str(ssh_dir)])
-        run_local(["chmod", "600", str(ssh_dir / "known_hosts")])
-        for ip in self.node_ips.values():
-            run_local(["ssh-keygen", "-R", ip], check=False)
-            run_local(f"ssh-keyscan -H {shlex.quote(ip)} >> {shlex.quote(str(ssh_dir / 'known_hosts'))}", check=False)
+        pass
 
     def prepare_remote_nix(self, ip):
         self.remote(ip, "sudo mkdir -p /etc/nix")