increase NeuVector limits and requests to avoid OOMKilled

The following discussion from !8211 (merged) should be addressed:

There is an OOMKilled status on the Neuvector controller pod.

See logs:

status:
  conditions:
  - lastProbeTime: null
    lastTransitionTime: "2026-06-15T10:14:58Z"
    status: "True"
    type: PodReadyToStartContainers
  - lastProbeTime: null
    lastTransitionTime: "2026-06-15T10:14:32Z"
    status: "True"
    type: Initialized
  - lastProbeTime: null
    lastTransitionTime: "2026-06-15T10:23:35Z"
    message: 'containers with unready status: [neuvector-controller-pod]'
    reason: ContainersNotReady
    status: "False"
    type: Ready
  - lastProbeTime: null
    lastTransitionTime: "2026-06-15T10:23:35Z"
    message: 'containers with unready status: [neuvector-controller-pod]'
    reason: ContainersNotReady
    status: "False"
    type: ContainersReady
  - lastProbeTime: null
    lastTransitionTime: "2026-06-15T10:14:32Z"
    status: "True"
    type: PodScheduled
  containerStatuses:
  - allocatedResources:
      cpu: 100m
      memory: 2280Mi
    containerID: containerd://6c7c0c3531d5f732f3631bb3c62b5d60b8c32d4267f6ddfb5b8116847412774f
    image: docker.io/neuvector/controller:5.4.1
    imageID: docker.io/neuvector/controller@sha256:bcc59e6fed67ccb03fb5db673df2c595d174656c84e31e6ad5be490be0b2e19e
    lastState:
      terminated:
        containerID: containerd://2707c6a6690391c367896fbaa340c2e6acc06384360d69f071da34360a7084a9
        exitCode: 137
        finishedAt: "2026-06-15T10:23:35Z"
        reason: OOMKilled
        startedAt: "2026-06-15T10:21:00Z"
    name: neuvector-controller-pod
    ready: false
    resources:
      limits:
        cpu: 400m
        memory: 2792Mi
      requests:
        cpu: 100m
        memory: 2280Mi
    restartCount: 3
    started: true
    state:
      running:
        startedAt: "2026-06-15T10:24:02Z"

We probably needs to increase the memory limit? For instance 4096Mi?

@tmmorin if we have to increase the limit, let's do it

but then we also need to increase the requests to let the scheduler account for its likely usage when deciding where to schedule it

and we need to choose values after observing what neuvector uses in actual deployments