otlpgrpc: turn on round_robin LB policy and kuberesolver for resolving k8s service endpoints

[yeya24]

What this PR does:

For tracing with opentelemetry, Cortex uses otlpgrpc protocol to send traces to OTLP backends.

When you have multiple tracing backends, it is always recommended to have load balancing. Since gRPC is a long lived connection, it is more recommended to do client side loadbalancing. For more details, can refer to https://github.com/grpc/grpc/blob/master/doc/load-balancing.md. In this pr, I added an option to enable round_robin policy, by default it is pick_first.

This pr also enables kuberesolver support which allows to resolve endpoint with prefix kubernetes://, for example kubernetes://opentelemetry-collector.tracing:4317 to be resolved as a k8s service and return the actual endpoint IPs (pod IP). This is helpful because kuberesolver watches the backend endpoints changes and allows client side loadbalancing to send requests to newly scaled up pods such as through HPA.

Which issue(s) this PR fixes:
Fixes #

Checklist

• Tests updated
• Documentation added
• CHANGELOG.md updated - the order of entries should be [CHANGE], [FEATURE], [ENHANCEMENT], [BUGFIX]