Call helm uninstall with --wait, and where necessary, wait for viking
service pods to be undeployed.

This minimizes errors during the sequential e2e test run -- the
controller may still be working on deletion of the resources of the
previous test when resources of the next test are added.