Change the log level and the message when ECS client cannot resolve public IPs.

[enozcan]

Per discussion on #218 (comment), ECS client logs a warn when public IPs are not fetched. Also from the previous comment of Rafal:

• "It should not crash the whole Hazelcast" - It already does not.

• "It should not print retry logs (at the INFO level)" - Retry logs are always printed at fine level:
  

  hazelcast-aws/src/main/java/com/hazelcast/aws/RetryUtils.java

  Lines 57 to 58 in e78d8c2

  	LOGGER.fine(String.format("Couldn't connect to the AWS service, [%s] retrying in %s seconds...",
  	retryCount, waitIntervalMs / MS_IN_SECOND));

• "It should only print a warning...We treat it as a correct configuration" - Changed with this PR and it behaves like that now.

This is not the case for EC2 as describeInstance permission already contains this public ip data.

The PR also makes the behavior consistent with the Kubernetes one.

[leszko]

👍