fix(dra): support Deckhouse GPU attributes

[danilrwx]

What this PR does

Before this PR:

The KubeVirt DRA status controller only recognized the upstream Kubernetes GPU device attributes (resource.kubernetes.io/pcieRoot, resource.kubernetes.io/mDevUUID). A GPU published by the Deckhouse GPU DRA driver under its own namespaced attributes (gpu.deckhouse.io/pciAddress, gpu.deckhouse.io/deviceType, gpu.deckhouse.io/sharingStrategy) could not be resolved into a GPU status, so VM GPU passthrough based on the Deckhouse driver did not work.

After this PR:

The DRA status controller understands the Deckhouse GPU attributes in addition to the upstream ones, and validates that an allocated GPU is safe for VM passthrough:

• reads gpu.deckhouse.io/pciAddress, gpu.deckhouse.io/deviceType (physical/mig), and gpu.deckhouse.io/sharingStrategy;
• rejects GPUs that allowMultipleAllocations, use any sharing strategy, are MIG without an mDevUUID, or expose a PCI address with a non-physical device type;
• normalizes 8-hex-digit PCI domain prefixes to the 4-digit form KubeVirt expects.

References

• Required by feat(vm): add GPU DRA resource claim support virtualization#2520 (VM GPU DRA support).

Why we need it and why it was done in this way

The Deckhouse GPU DRA driver publishes device attributes under the gpu.deckhouse.io/ namespace rather than resource.kubernetes.io/. The virtualization module selects GPUs by productName and only accepts exclusive physical devices, so KubeVirt must read the Deckhouse attributes and enforce the same exclusivity constraints as a safety layer.

The following tradeoffs were made:

• Exclusivity validation lives in KubeVirt as the last safety layer. The virtualization controller also constrains selection with a CEL selector, so the constraints are enforced in two places intentionally.

The following alternatives were considered:

• Rename the Deckhouse driver attributes to the upstream resource.kubernetes.io/ keys — rejected to keep driver-specific metadata namespaced and avoid collisions with other DRA drivers.

Links to places where the discussion took place: deckhouse/virtualization#2520

Special notes for your reviewer

Deckhouse-only downstream change. Tracked under the deckhouse/virtualization GPU DRA effort. The unrelated chore(lint): migrate golangci config to v2 commit is bundled on this branch.

Checklist

• Design: A design document was considered and is present (link) or not required
• PR: The PR description is expressive enough and will help future contributors
• Code: Write code that humans can understand and Keep it simple
• Refactor: You have left the code cleaner than you found it (Boy Scout Rule)
• Upgrade: Impact of this change on upgrade flows was considered and addressed if required
• Testing: New code requires new unit tests. New features and bug fixes require at least one e2e test
• Documentation: A user-guide update was considered and is present (link) or not required. You want a user-guide update if this is a user facing feature / API change.
• Community: Announcement to kubevirt-dev was considered

Release note

DRA: recognize Deckhouse GPU device attributes (gpu.deckhouse.io/pciAddress, deviceType, sharingStrategy) and reject GPUs unsuitable for VM passthrough (multi-allocation, shared, MIG without mdevUUID, non-physical with PCI address).