fix: configure kubectl before cleanup in make destroy

[anners]

Closes #26

Problem

make destroy called make clean which runs kubectl delete before kubectl was configured for the cluster. This caused the destroy to fail if kubectl wasn't already pointing at the cluster.

Fix

Replaced make clean in the destroy target with an inline step that:

1. Configures kubectl via terraform output -raw configure_kubectl
2. Deletes the test workload
3. If either step fails (cluster already unreachable or partially destroyed), prints a message and continues rather than blocking the Terraform destroy