KAFKA-19117: Client Throttling Log messages should be of log level - WARN (Java client)

[siddharthaDevineni]

We experienced an outage in our application due to a built up Kafka lag, which we eventually discovered to be the problem with Kafka Clients being throttled.

Even when using the confluent cluster, the dashboard does point to exceeding the quota of the principal.

It will be nice, if the log message on the client side is a WARNING. This helps those who don't have visibility to cluster, and just look at the problem from the client side, setup alerts etc. (instead of enabling TRACE to figure out where the problem is)

NetworkClient#maybeThrottle in the class NetworkClient:

Changed from:
log.trace("Connection to node {} is throttled for {} ms until timestamp {}", nodeId, throttleTimeMs, now + throttleTimeMs);

to

log.warn("Connection to node {} is throttled for {} ms until timestamp {}", nodeId, throttleTimeMs, now + throttleTimeMs);

[github-actions]

A label of 'needs-attention' was automatically added to this PR in order to raise the
attention of the committers. Once this issue has been triaged, the triage label
should be removed to prevent this automation from happening again.