Batch otel logs and disable daemon cluster metrics

[ArnobKumarSaha]

• Gateway logs pipeline had no batch processor, so every small log chunk became its own ClickHouse INSERT (~2.5 inserts/s of ~7 rows on telecash-dbaas). This caused constant part creation and merges. Add batch/logs (5000 / 10000 / 5s).
• The daemon inherited presets.clusterMetrics: true from the opentelemetry-kube-stack defaults. k8s_cluster metrics carry pod/namespace only as resource attributes, and prometheusremotewrite drops those, so every k8s_* metric collapsed to one series. Thanos then rejected the rest with duplicate sample for timestamp (~40 errors/min). kube-state-metrics already provides these metrics, so disable the preset.

---

^{Need help on this PR? Tag @codesmith-bot with what you need. Autofix is disabled.}