dra: fix HOST_ROOT envvar and add missing mounts for PassthroughSupport

[cdesiniotis]

Description

This PR fixes a few bugs encountered when enabling the PassthroughSupport feature gate in the NVIDIA DRA Driver for GPUS. First, the HOST_ROOT variable was not set correctly, leading to the following error message in the kubelet-plugin:

Error: invalid CLI flags: host root is not mounted at "/host-root"

After this issue was fixed, the below issue was encountered as we were missing some volume mounts required for the PassthroughSupport feature.

Error: error creating driver: error enumerating all possible devices: error enumerating GPU PCI devices: error getting GPU info from PCI device: error finding best VFIO driver for device "0009:01:00.0": failed to get vfio_pci aliases in modules.alias file: failed to read file /lib/modules/5.14.0-687.20.1.el9_8.aarch64/modules.alias: open /lib/modules/5.14.0-687.20.1.el9_8.aarch64/modules.alias: no such file or directory

See upstream DRA driver manifest: https://github.com/kubernetes-sigs/dra-driver-nvidia-gpu/blob/50e0e3568f633d87266e3aabf8afa510843d6c7d/deployments/helm/dra-driver-nvidia-gpu/templates/kubeletplugin.yaml#L338-L353

Checklist

• No secrets, sensitive information, or unrelated changes
• Lint checks passing (make lint)
• Generated assets in-sync (make validate-generated-assets)
• Go mod artifacts in-sync (make validate-modules)
• Test cases are added for new code paths

Testing

Manual testing on a single node OpenShift cluster. Deployed with and without this change.

[tariq1890]

LGTM. Once gofmt is applied, this should be good to merge.

[cdesiniotis]

LGTM. Once gofmt is applied, this should be good to merge.

Fixed now.