E2e test planning

[alicefr]

Let's discuss about the test planning. Currently, we only test the happy path for the reboot

Ideas about follow-up tests:

1. Extend the test TestUpdateReboot by rolling back to the previous image
2. Pause the rollout and check that the resuming continue correctly
3. Try to upgrade to no-existing image, the node should go in degraded state and the rollout should be paused
4. Shutdown (manually) a node after the reboot to simulate a faulty node, after the upgrade. It should be reported as degraded
5. Try to kill the controller/daemon during a roll-out. They should recover and the rollout should progress and terminate successfully
6. Testing the happy path with the digests (once supported)
7. Upgrade to an image protected by a pull secret (once supported)
8. Try to switch to a completely different image e.g fedora to centos stream
9. Test upgrades for the operator

[jlebon]

One general note: since each test spawns a separate worker node, which takes time, I would bundle these subtests into tests where it makes sense. E.g. error conditions/recovery tests could all be checked together in one test for example.

[jlebon]

Try to switch to a completely different image e.g fedora to centos stream

Is the more general thing we're testing here that the node degrades successfully if rebooting into an image which doesn't actually have e.g. the kubelet?

Test upgrades for the operator

This relates to #14.