We've seen intermitent failures on `delete-workload-cluster`: ``` /bin/bash: line 681: bin/env: No such file or directory ``` This seems to only impact ~capo pipelines and seems to be due to the changes we made on gitlab CI file cache (according to @jonathang the cache is per-runner-slot and now that it's also per dev branch, it's become unlikely to have the toolbox be found in cache; and unfortunately, it seems that until now the assumption of delete-workload-cluster has been that the toolbox would be there.