Selected context not respected when checking if namespace exists

[rootmos]

During startup when checking if the selected namespace exists, the context is not supplied to the kubectl get ns command:

telepresence/telepresence/startup.py

Line 127 in c6a5e56

runner.get_output([prelim_command, "get", "ns",

That is, it'll use your global context, which not necessarily is the desired context. Reproduce by setting the global context to a cluster without the selected namespace.

I propose the following fix (read: works for me, was broken before): toknapp@571a862

[kooba]

This issue still exists. Workflows that supply both --context and --namespace and don't have current-context set in kube/config do not work at the moment and error T: Error: Namespace 'mynamespace' does not exist is raised.