fix inconsistent behavior of running kube-proxy on 1.12

[Huang-Wei]

What this PR does / why we need it:

As feature ScheduleDaemonSetPods will be promoted to beta in 1.12, it will result in a "Pending" kube-proxy pod for a fresh kubeadm installation. PS: upgrade case should not be impacted.

This PR is to make sure kubeadm behaviors the same in 1.12 on kube-proxy daemonset.

Which issue(s) this PR fixes:
Fixes part of #66831

Special notes for your reviewer:

Release note:

kube-proxy is now deployed using nodeAffinity with explicit matchExpressions.

[k8s-ci-robot]

[APPROVALNOTIFIER] This PR is NOT APPROVED

This pull-request has been approved by: Huang-Wei
To fully approve this pull request, please assign additional approvers.
We suggest the following additional approver: fabriziopandini

If they are not already assigned, you can assign the PR to them by writing /assign @fabriziopandini in a comment when ready.

The full list of commands accepted by this bot can be found here.

The pull request process is described here

Needs approval from an approver in each of these files:

• cmd/kubeadm/OWNERS

Approvers can indicate their approval by writing /approve in a comment
Approvers can cancel approval by writing /approve cancel in a comment

[Huang-Wei]

/sig cluster-lifecycle

[MHBauer]

/ok-to-test

[neolit123]

As feature ScheduleDaemonSetPods will be promoted to beta in 1.12

but it becomes a question if we want to enable it by default in kubeadm in 1.12.

/assign @luxas
/assign @timothysc

[Huang-Wei]

/hold

@neolit123 we're evaluating if it's better to give a fix so as to keep the behavior consistent

[timothysc]

... Why is the previous node-selector inconsistent?

Affinity is super expensive.

/cc @bsalamat

[Huang-Wei]

@timothysc sorry for the late /hold label - we're internally evaluating changes to keep the behavior consistent. Most probably we will go that way, and then we don't need this change and will close this pr.

[bsalamat]

@timothysc as @Huang-Wei said already, we probably won't need this change. We will fix the code on the DS controller side so that the old configuration keeps working.
BTW, inter-pod affinity is expensive. Node affinity which is used here is not that expensive.

[luxas]

And long term we're hoping we can completely ditch this so we can just use manifest lists.
cc @dims @neolit123 @mkumatag

[Huang-Wei]

Current kube-proxy manifest will continue to work due to #66953.

/close