fix: retry ExceptionHandler not retrying on IOException

[PhongChuong]

This PR fix an issue where the retry algorithm did not retrying based on the BIGQUERY_EXCEPTION_HANDLER configuration. This is caused by HttpBigQueryRpc translating IOExceptions into BigQueryException does not match the configurations set by BIGQUERY_EXCEPTION_HANDLER. To resolve this, we implement new internal methods (.*SkipExceptionTranslation) in HttpBigQueryRpc which do not translate the IOException. All retryable calls to HttpBigQueryRpc are updated to use these new methods.

In addition,

• To preserve existing BigQueryException structure, this PR updates BigQueryRetryHelper.runWithRetries to parse the IOExceptions thrown by the .*SkipExceptionTranslation methods and translate them into BigQueryRetryException. This is needed as the original HttpBigQueryRpc translate IOExceptions into BigQueryException.
• To preserve return values in the BigQuery interface for special cases such as HTTP_NOT_FOUND, this PR updates BigQueryImpl to checks the error code and return the appropriate BigQueryException.
• The unit test BigQueryImplTest.testCreateJobFailureShouldRetryExceptionHandlerExceptions verify that BIGQUERY_EXCEPTION_HANDLER exceptions (UnknownHostException & ConnectException) are being retried

Fixes b/394167052 ☕️

[whuffman36]

LGTM! Nice work on such a massive PR

[shollyman]

One more minor callout for delete behavior, but otherwise LGTM.

Also, "SkipExceptionTranslation" may forever be burned into my retinas, but thanks for slogging through this refactor.

[PhongChuong]

Thanks for the review. I've comment on the isRetryErrorCodeHttpNotFound check variation for delete job. PTAL