[P0] event-exporter: migrate watcher to events.k8s.io/v1 to reduce event churn

## Problem
`event-exporter` watches `core/v1` Events. These objects are updated frequently (count/lastTimestamp), generating high update volume and write amplification to Cloud Logging under heavy load. This increases CPU, memory, and egress in large clusters.

## Proposed optimization
Switch the event watcher to `events.k8s.io/v1` and map `EventSeries` to reduce update frequency. In the newer Events API, repeated events are represented as a series (`series.count`, `series.lastObservedTime`) rather than continuous updates to a single core/v1 Event. This should significantly reduce update traffic and improve throughput in high-volume clusters.

## Notes / references
- Events API group: https://kubernetes.io/docs/reference/kubernetes-api/cluster-resources/event-v1/
- EventSeries semantics: https://kubernetes.io/docs/reference/kubernetes-api/cluster-resources/event-v1/#EventSeries
- client-go typed client: `k8s.io/client-go/kubernetes/typed/events/v1`

## Acceptance criteria
- Watcher uses `events.k8s.io/v1` client and handles `EventSeries` semantics.
- Observed reduction in update volume vs. `core/v1` (documented via metrics or basic benchmarking).
- Backward-compatible behavior for log entry fields used by sink (no loss of required data).