kueue: Add Topology-Aware Scheduling resource model and views

[abhayrajjais01]

Description

This PR introduces support for Kueue's Topology CRD (kueue.x-k8s.io/v1alpha1 / kueue.x-k8s.io/v1beta1) in the Headlamp Kueue plugin.

Topology-Aware Scheduling (TAS) is a feature in Kueue that allows cluster administrators to model physical hardware topology (such as rack -> block -> node -> gpu) so that large-scale distributed AI/ML training workloads (Ray, PyTorch, vLLM) are co-located within low-latency interconnect boundaries.

What Changes

Resource Model & Formatters (src/resources/):
Implemented Topology model extending KubeObject<KubeTopology>.
Added pure formatters for level hierarchy chains (renderTopologyLevelsSummary), level counts (renderTopologyLevelsCount), and node label associations.

UI Views (src/components/topologies/):
TopologyList: ResourceListView displaying name, levels hierarchy (rack (topology.kubernetes.io/rack) → block → node), total levels, and age with useTranslation() localization.
TopologyDetail: DetailsGrid showing total levels, level chain, and an interactive Topology Levels Hierarchy card inspector detailing level numbers, names, and node labels.

Navigation & Routing:
Registered Topologies sidebar navigation entry under the Kueue menu group.
Registered list (/kueue/topologies) and detail (/kueue/topologies/:name) routes in kueueRoutes.ts and index.tsx.

Unit Tests (src/resources/topology.test.ts):
Added 8 unit tests covering level formatting, null fallbacks, arrow chain rendering, and level extraction.

How Has This Been Tested?

• Ran unit tests with Vitest: 5/5 test files passed, 30/30 unit tests passed.

Checklist

• Follows Headlamp commit guidelines
• DCO signed-off commit
• Includes unit tests