Add sleep to helm test

In CI and locally we are noticing that the helm script test fails to complete with a connection refused on the curl commands. This seems to be some odd function of the pod startup (potentially a delay in the "allow" networkpolicies applying to the pod). Running a standalone curl pod works, but only if you start the pod and then run the curl about 10-15 seconds after.

A lazy/simple solution is provided to give the pod some time to wake up.