ParallelIterable.close() will not clear queue

[Heltman]

Apache Iceberg version

0.14.1 (latest release)

Query engine

Trino

Please describe the bug 🐞

When Trino query Iceberg failure in file scan time, it call fileScanTaskIterable.close(). However, because the queue in ParallelIterable is not cleared, it will occupy more Coordinator memory, and even cause OOM.

This is a question related to #4596 , this queue may be use too much memory in coordinator when query run. But when query failed, this queue still use much memory until trino expire this query(default: 15min and newest 100 query).

We need to confirm that the queue has been cleared when close() method is called.

[rdblue]

@Heltman, would you like to open a PR to clear the queue?

[Heltman]

@rdblue Yes, I had