Schema validation already happens in prepareRun(), so we don't need it in initialize().
As part of validation, we call BigQuery.getTable(). These calls can be rate limited if they are called concurrently from many exevcutors. We can avoid rate limiting issues by validating schema only in prepareRun.
Removed redundant validations from BQ sink , this should reduce calls to BigQuery.getTable()
Fix available on Jun 25th RC build