event-exporter: prefer events.k8s.io/v1 + add e2e harness

[erain]

Summary

This change migrates the event-exporter watcher to prefer events.k8s.io/v1 while preserving the existing core/v1 sink pipeline. It adds an auto API selection mode (default) that uses discovery to pick the best API and safely falls back to core/v1 when events.k8s.io/v1 is unavailable. It also adds a reusable local E2E harness (kind + build-tagged test), plus unit/integration tests and micro-benchmarks so we can validate parity and capture performance improvements.

Fixes #1028.

Key changes

• API selection + fallback: new --events-api flag (default auto) resolves to events.k8s.io/v1 when available, otherwise core/v1.
• Conversion path: events.k8s.io/v1 objects are converted to core/v1 in the watcher handler so the sink, log entry factory, and monitored resource mapping remain unchanged.
• Timestamp robustness: log entry timestamps now fall back to EventTime when LastTimestamp/Series are missing.
• E2E workflow: build-tagged E2E test + hack/e2e-kind.sh + make e2e target, documented in event-exporter/docs/e2e.md.
• Tests + bench: new unit tests for conversion and API resolution, integration test for v1 watch delivery, and benchmarks for log entry processing.

Acceptance criteria (from #1028)

• Watcher uses events.k8s.io/v1 + EventSeries: implemented watcher path with conversion and unit/integration coverage.
• Reduction in update volume: E2E results show fewer processed updates and higher throughput for events/v1 (see Performance section).
• Backward-compatible sink fields: conversion keeps core/v1 sink pipeline; log entry fields validated by tests and E2E samples.

Testing

• go test ./... (from event-exporter/)
• go test ./sinks/stackdriver -bench BenchmarkProcess -benchmem
• make e2e (kind)

Performance (local kind E2E)

Environment:

KIND_CLUSTER_NAME=event-exporter-e2e
E2E_EVENT_COUNT=200
E2E_SERIES_STEP=10
E2E_TIMEOUT=2m
E2E_NAMESPACE=event-exporter-e2e

Results:

• core/v1: processed=200, updates=0, occurrences=200, elapsed=38.59s, throughput=5.18 occ/s
• events.k8s.io/v1: processed=21, updates=20, occurrences=200, elapsed=2.40s, throughput=83.39 occ/s

Benchmarks (log entry factory)

• BenchmarkProcessCoreV1RepeatedEvents: ~4.23ms/op, ~492 KB/op, ~9.8k allocs/op (100 occurrences)
• BenchmarkProcessSeriesAggregatedEvent: ~0.052ms/op, ~5.7 KB/op, ~116 allocs/op (100 occurrences)

Notes

• event-exporter/go.mod gains github.com/spf13/pflag as an indirect dependency because the E2E test uses clientcmd for kubeconfig loading.

[google-cla]

Thanks for your pull request! It looks like this may be your first contribution to a Google open source project. Before we can look at your pull request, you'll need to sign a Contributor License Agreement (CLA).

View this failed invocation of the CLA check for more information.

For the most up to date status, view the checks section at the bottom of the pull request.