Search before asking

• I searched in the issues and found nothing similar.

Motivation

Although #954 already introduced timeout closableiterator to fast fail flink test if it get more times to collect records before it reach to the ci 60min timeout. But we still see the test stuck util ci 60min timeout.
Here is just one example:
https://github.com/apache/fluss/actions/runs/16648794360

Solution

• Correct the timeout closableiterator. Currently, it won't be timeout, since flink closableiterator#hasNext() will always block utils records comes in streaming mode
• Make all directly use of flink closableiterator.next to use the timeout collect method

Anything else?

No response

Willingness to contribute

• I'm willing to submit a PR!