Re-enable GPU profiles for GCP/AWS

[viniciusdc]

Fixes | Closes | Resolves #1209

Changes introduced in this PR:

• Created a new file on 03-kubernetes-initialize stage, for Nvidia driver daemonset
• Included new import var gpu_node_group_names to validate count for daemonset resource
• Added schema validation for guest_accelerators

Types of changes

What types of changes does your PR introduce?

Put an x in the boxes that apply

• Bug fix (non-breaking change which fixes an issue)
• New feature (non-breaking change which adds a feature)
• Breaking change (fix or feature that would cause existing features to not work as expected)
• Documentation Update
• Code style update (formatting, renaming)
• Refactoring (no functional changes, no API changes)
• Build related changes
• Other (please describe):

Testing

Requires testing

• Yes
• No

In case you checked yes, did you write tests?

• Yes
• No

Further comments (optional)

Tested the GPU was available in the profile executing nvidia-smi

[viniciusdc]

This re-enables Profiles with GPU support, I think we should also include in this PR the jupyterhub version bump, as the GPU global settings issue #1201 might show during unexpected circumstances:

│   Normal   NotTriggerScaleUp  3m51s                cluster-autoscaler  pod didn't trigger scale-up: 1 node(s) had taint {nvidia.com/gpu: present}, that the pod didn't tolerate, 2 node(s) didn't match Pod's node affinity/selector, 1 m │
│ ax node group size reached                                                                                                                                                                                                                │
│   Warning  FailedScheduling   39s (x5 over 3m53s)  default-scheduler   0/3 nodes are available: 1 node(s) didn't match Pod's node affinity/selector, 1 node(s) had taint {node.kubernetes.io/memory-pressure: }, that the pod didn't tole │
│ rate, 1 node(s) had taint {nvidia.com/gpu: present}, that the pod didn't tolerate.     

Those messages was originated from the conda-store worker pod after it was evicted

[viniciusdc]

The reason I opted for

variable "gpu_node_group_names" {
  description = "Names of node groups with GPU"
  default = []
}

instead of a simples bool conditional to enable the daemon was to keep a standard to what was done with AWS provider. We can change that later if needed.

[viniciusdc]

This should not be merged until:

• The post-release is completed
• The taint issue above, nvidia.com/gpu: present, is fixed within this PR (or a linked PR is submitted)

[viniciusdc]

In hold for #1227

[viniciusdc]

Just found a bug, pushing the fix right away

[viniciusdc]

In hold for quota increase to test AWS