fix #3190 only use nvidia-container-runtime on gpu nodes

[johnhofman]

What this PR does / why we need it:
Allows deployment of a cluster with cpu and gpu agent pools with the nvidia-device-plugin enabled. Without this the docker daemon on the cpu nodes fails to start.

Which issue this PR fixes *
This fixes #3190 by only making the changes to /etc/docker/daemon.json on gpu enabled nodes using the same template switch which enables/disables the installation of the GPU drivers.

[acs-bot]

[APPROVALNOTIFIER] This PR is NOT APPROVED

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

Assign the PR to them by writing /assign @jackfrancis 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:

• OWNERS

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

[sozercan]

Thanks @johnhofman!

#3181 also addresses this issue and adds toleration for device plugin so it doesn't get scheduled on CPU nodes. I'll close this in favor of #3181 unless you see anything else missing?