[Bug]: [Java BQ] Reduce the number of retries for streaming Storage API writes

### What happened?

Streaming Storage Write API has a RetryManager with up to 1000 retries and a back-off restricted between 1-10 seconds:
https://github.com/apache/beam/blob/7eaef18bcbfb48fb112884564b1362f638fe5834/sdks/java/io/google-cloud-platform/src/main/java/org/apache/beam/sdk/io/gcp/bigquery/StorageApiWritesShardedRecords.java#L740-L741

Maybe we should revisit these decisions as 1000 retries might not be needed in a streaming pipeline. In the Dataflow case, if we fail earlier, the runner will retry the bundle as many times as needed. Also, are there any errors that warrant us waiting more than 10 seconds before retrying?

### Issue Priority

Priority: 2 (default / most bugs should be filed as P2)

### Issue Components

- [ ] Component: Python SDK
- [X] Component: Java SDK
- [ ] Component: Go SDK
- [ ] Component: Typescript SDK
- [X] Component: IO connector
- [ ] Component: Beam examples
- [ ] Component: Beam playground
- [ ] Component: Beam katas
- [ ] Component: Website
- [ ] Component: Spark Runner
- [ ] Component: Flink Runner
- [ ] Component: Samza Runner
- [ ] Component: Twister2 Runner
- [ ] Component: Hazelcast Jet Runner
- [ ] Component: Google Cloud Dataflow Runner