feat: Support Prometheus metrics for CSI and kubelet with Grafana dashboard

[sheyabernstein]

Summary

• Instruments CSI operations (attach, detach, node stage/unstage, publish/unpublish) with latency histograms, total counters, and error counters via named return + deferred ObserveCSIOperation
• Adds csi_volume_attachments_total gauge tracking volumes currently mounted per node, incremented/decremented in NodePublishVolume / NodeUnpublishVolume
• Implements NodeGetVolumeStats (was a stub) to return byte and inode usage, enabling kubelet_volume_stats_* metrics
• Adds Helm values for metrics.serviceMonitor (creates headless + ClusterIP Services and a ServiceMonitor CR) and metrics.dashboards (ConfigMap for Grafana sidecar)
• Grafana dashboard with CSI operation panels and a PVC usage table (capacity, used, available, LCD gauge for usage %)