fix: add pod UID to DRA-attributed metrics

[git-jxj]

Description

Honor KubernetesEnablePodUID in the DRA attribution path, matching the existing device-plugin path. DRA-attributed metrics now include the existing pod_uid attribute when the option is enabled.

The regression test exercises the full fake PodResources and DRA ResourceSlice mapping path for a MIG metric and verifies the UID alongside the existing pod and DRA attributes.

Testing

• go test ./internal/pkg/transformation -run "^TestPodMapperProcessAddsDRAAttributesForMIGMetric$" -count=1
• go test ./internal/pkg/transformation -count=1
• CPU-safe repository package sweep excluding tests/k8s, integration_test, and nvmlprovider
• make lint LINT_BASE_REV=origin/main
• make check-fmt

A direct go test ./... -short also reached the repository-wide packages; only the existing GPU/DCGM-backed packages failed because this host lacks libdcgm.so.4 and has no matching NVML driver.

Fixes #718

[git-jxj]

Current head 6b375ca has not run the Go workflow: the run is action_required pending outside-contributor approval. Could a maintainer approve this workflow run so CI can validate the change? https://github.com/NVIDIA/dcgm-exporter/actions/runs/34957252600

[nccurry]

This should be fixed in 4.8.4. Closing. Please re-open or open a new issue or PR if you find issues.