[java-storage] Batch operations do not use the RetryHelper settings

#### Stack trace
```
2026-02-25 15:43:42.660
at com.google.cloud.storage.StorageException.translate(StorageException.java:230) ~[bq-sink-batch.jar:0.1.0-SNAPSHOT]
2026-02-25 15:43:42.660
at com.google.cloud.storage.spi.v1.HttpStorageRpc.translate(HttpStorageRpc.java:360) ~[bq-sink-batch.jar:0.1.0-SNAPSHOT]
2026-02-25 15:43:42.660
at com.google.cloud.storage.spi.v1.HttpStorageRpc.access$500(HttpStorageRpc.java:119) ~[bq-sink-batch.jar:0.1.0-SNAPSHOT]
2026-02-25 15:43:42.660
at com.google.cloud.storage.spi.v1.HttpStorageRpc$DefaultRpcBatch.submit(HttpStorageRpc.java:330) ~[bq-sink-batch.jar:0.1.0-SNAPSHOT]
2026-02-25 15:43:42.660
at com.google.cloud.storage.StorageBatch.submit(StorageBatch.java:159) ~[bq-sink-batch.jar:0.1.0-SNAPSHOT]
2026-02-25 15:43:42.660
at com.google.cloud.storage.StorageImpl.delete(StorageImpl.java:1268) ~[bq-sink-batch.jar:0.1.0-SNAPSHOT]
2026-02-25 15:43:42.660
at org.apache.flink.fs.gs.storage.GSBlobStorageImpl.delete(GSBlobStorageImpl.java:246) ~[?:?]
2026-02-25 15:43:42.660
at org.apache.flink.fs.gs.writer.GSRecoverableWriterCommitter.cleanupTemporaryBlobs(GSRecoverableWriterCommitter.java:244) ~[?:?]
2026-02-25 15:43:42.660
at
```

#### External references such as API reference guides

- https://github.com/googleapis/java-storage/issues/203

#### Any additional information below

https://github.com/googleapis/java-storage/issues/203

Hi, I wanted to reopen issue 203 as we saw it come up in our Flink application. Despite adjusting all of our other retry settings, we noticed that delete batch requests were not being retried after tracing through the code and finding the previous issue opened.

Each of these failures can cause a cascading effect when GCP buckets are under strain where that Flink checkpoint fails, and the subsequent one will submit more delete batch requests when catching up which are then more likely to run into this terminal error.

It's really confusing as a library user that the normal HTTP retry settings have no effect here so I would petition that we reopen and use https://github.com/googleapis/java-storage/pull/305

Start by tracing the batch request path shown in the stack trace, from StorageImpl.delete through StorageBatch.submit and HttpStorageRpc$DefaultRpcBatch.submit, and compare it with the RetryHelper settings described in the issue. Review the linked java-storage issue and pull request 305; done means batch delete requests honor the configured retry settings and the behavior is covered by appropriate tests.

由索引模型根据 Issue 内容生成。