What this PR does / why we need it:

Summary

• Adds the pod-level resourceClaims field ([]corev1.PodResourceClaim) to the FeatureStoreServices struct, enabling Dynamic Resource Allocation (DRA) support for FeatureStore deployments.
• Adds controller plumbing (applyResourceClaims) to propagate the field into the Deployment's PodSpec during reconciliation.
• Regenerates CRD manifests and DeepCopy functions.
• Adds unit tests covering field propagation, nil/empty cases, and removal.

Background

Kubernetes Dynamic Resource Allocation (DRA) uses a claim-based model for hardware resources like GPUs. The container-level resources.claims path was already available for each Feast service (via OptionalCtrConfigs.Resources), but the pod-level resourceClaims field — which defines and allocates the ResourceClaim objects — was missing. Without it, container-level claims have nothing to reference.

This change completes the DRA chain:

• Pod-level resourceClaims → allocates the claim
• Container-level resources.claims → consumes the allocated claim

Example FeatureStore CR

apiVersion: feast.dev/v1
kind: FeatureStore
metadata:
  name: my-feast
spec:
  feastProject: my_project
  services:
    # Pod-level: define claims available to any container in the pod
    resourceClaims:
      - name: gpu-claim
        resourceClaimName: my-gpu-resource-claim

    # Online store container consumes the GPU claim
    onlineStore:
      server:
        resources:
          claims:
            - name: gpu-claim

    # Registry does not need the GPU
    registry:
      local:
        server: {}