pre-stop failure when kubeconfig is configured as a path-config-file

[poblin-orange]

When helm-kubectl bosh release is used with another bosh release, creating the kubeconfig as a path-config-file (typically k3s bosh release), the pre-stop script breaks and locks the deployment when the kubeconfig.yml is not found.

https://github.com/orange-cloudfoundry/helm-kubectl-boshrelease/blob/master/jobs/action/spec#L36

This can happen at first deployment time (order of pre-start/pre-stop is not predictible in bosh)
=> the pre-stop should not depend on kubeconfig.yml file existence to check the target bosh deployment state.

https://github.com/orange-cloudfoundry/helm-kubectl-boshrelease/blob/master/jobs/action/templates/bin/pre-stop.erb#L2

    server/04cacacf-fbe4-46bf-bb44-2c359d512daf:/var/vcap/jobs/action/bin# ./pre-stop 
    found correct rubygems version 3.1.2
    cp: cannot stat '/var/vcap/store/k3s-server/kubeconfig.yml': No such file or directory

[poblin-orange]

still issues with pre-stop. Should disable this feature (not usefull in a general way, dangerous with multi master architecture)