Machines, ApprovedImages stay behind on namespace deletion

[Jakob-Naucke]

1. Run a test that fails, e.g. artificially:

diff --git a/tests/attestation.rs b/tests/attestation.rs
index 118a1de..4e300e5 100644
--- a/tests/attestation.rs
+++ b/tests/attestation.rs
@@ -76,2 +76,3 @@ async fn test_attestation() -> anyhow::Result<()> {
     test_ctx.info("Attestation successful: encrypted root device verified");
+    assert!(false, "testing");
     att_ctx.cleanup().await?;

2. Delete namespace:

% kubectl delete ns test-2ae567d7
namespace "test-2ae567d7" deleted
[hangs]

3. Resources are left behind:

% kubectl get all,cm,secret,machines,approvedimages
NAME                                             TYPE     DATA   AGE
secret/ak-24becb59-206c-47f0-84e1-e6935278c719   Opaque   1      12m

NAME                                                                                 AGE
machine.trusted-execution-clusters.io/machine-966666bf-6a34-4a1f-9fb0-b3e3b000d35f   13m

NAME                                                 AGE
approvedimage.trusted-execution-clusters.io/coreos   14m

[Jakob-Naucke]

and deleting these resources also hangs