e2e: CI fails in sequential installation tests

[ldevulder]

The failing tests run on RancherManager v2.7.6.

Resume of tests:

• GH build based on SLE Micro for Rancher (regular CI) ❌
• OBS_Dev build ✅
• OBS_Staging build ✅
• GH build based on Leap 15.4 ❌

And weirdly no issue with tests on Rancher Manager HEAD version:

• GH build on SLE Micro for Rancher (regular CI) ✅

Here some logs when it works: https://github.com/rancher/elemental/suites/15950137856/artifacts/910657747.
Here some logs when it fails (from the same test where the screenshot were made): https://github.com/rancher/elemental/suites/15959129746/artifacts/911235902.

When it fails I can see that when the 2nd node is added in the cluster the state of the 1st node is changed to Waiting for plan to be applied.

It can be easily reproduce in the CI and an environment for live debugging can be provisioned.

Here some screenshots when it fails:

[ldevulder]

Fix by this PR: rancher/elemental-operator#526.