Remove CPU limits from otel collectors

[ArnobKumarSaha]

Drop `resources.limits.cpu` from the agent and gateway collectors to avoid CFS throttling during scrape and remote-write bursts. Memory limits stay because `memory_limiter` uses `limit_percentage` of the container memory limit.

The subchart defaults are also removed: `defaultCRConfig` is deep-merged under every collector in `templates/collector.yaml`, and `collectors.daemon` is coalesced under the parent values. Removing only the parent values would render 200m/250m CPU limits. `cpu: null` is rejected by the subchart values schema.

Also sets resources where none were set, without CPU limits:

• gateway: memory 768Mi request / 2Gi limit. `memory_limiter` soft limit is 55% of the limit, so 1Gi started refusing data at ~560MB while retry queues grow during a hub outage.
• target allocator: 50m / 128Mi request, 512Mi limit (observed 2m / 75Mi).
• operator manager: 50m / 128Mi request, 512Mi limit (observed 1m / 33Mi).
• kube-rbac-proxy: 5m / 32Mi request, 128Mi limit (observed 0m / 10Mi).

Rendered:
```
x-gateway limits: {memory: 2Gi} requests: {cpu: 200m, memory: 768Mi}
x-agent limits: {memory: 2Gi} requests: {cpu: 200m, memory: 1Gi}
x-agent targetAllocator limits: {memory: 512Mi} requests: {cpu: 50m, memory: 128Mi}
operator manager limits: {memory: 512Mi} requests: {cpu: 50m, memory: 128Mi}
operator kube-rbac-proxy limits: {memory: 128Mi} requests: {cpu: 5m, memory: 32Mi}
```