vSphere Kube-up: resolve vm-names on all nodes.

[kerneltime]

What this PR does / why we need it:
This addresses the #35364

Which issue this PR fixes (optional, in fixes #<issue number>(, #<issue_number>, ...) format, will close that issue when PR gets merged): fixes #
fixes #35364
Special notes for your reviewer:
Will cherry pick to release-1.4

vSphere Kube-up: resolve vm-names on all nodes

---

This change is

[kerneltime]

Other than the issue being fixed there is no need for release-node-label-needed.

[eparis]

Do we do this on any other cloud providers? Or do we just expect that to be part of the infrastructure for things like GCE and AWS?

[kerneltime]

A lot of other deployments use IP address directly when nodes connect to master(using host override), the current implementation of vSphere Cloud Provider needs vm-names and we need this PR to resolve the vm-names to ip when dashboard retrieves logs.

Since kube-up is designed to be deployed on any kind of vSphere deployment, the VM ip addresses are discovered and the mapping information needs to be propagated.

Without this PR in 1.4 branch and master, any communication that is node to node fails and it includes getting logs in dashboard. This is not the same as pod to pod communication.

We are looking into changing how vSphere Cloud Provider manages vm-names and IPs and how we setup the nodes (support for kube-anywhere) but those changes will be further down the road and aimed at 1.5.

[k8s-github-robot]

Automatic merge from submit-queue