feat(kafka): add messaging.kafka.cluster.id to Kafka producer/consumer spans

Problem

Kafka spans (aiokafka, kafka-python, confluent-kafka) do not include messaging.kafka.cluster.id. When the same topic name exists across multiple clusters or environments, there is no way to distinguish which cluster a span came from.

Expected behavior

messaging.kafka.cluster.id (semconv) should be set on Kafka producer and consumer spans. It is a Recommended attribute in the Kafka messaging semantic conventions, so it is emitted by default (no opt-in flag), consistent with the other Recommended Kafka attributes these instrumentations already produce. The id is read from the client's own already-resolved metadata, so no extra broker connection is required.

Context

Part of a cross-language effort to add this attribute consistently across OTel instrumentations. See opentelemetry-java-instrumentation#18977 for the cross-language tracking issue.

Start by reading the aiokafka, kafka-python, and confluent-kafka instrumentation entry points and the existing Recommended Kafka span attributes. Check how each client exposes already-resolved metadata, then verify producer and consumer spans include messaging.kafka.cluster.id by default without an extra broker connection.