switch to exponential backoff instead of fixed interval

We are seeing issue with fixed interval retry: document count mismatched between ingestion and search service count.

This happens when Indexer service retry too soon on Storage service to fetch bulk records. Storage service responds with 404 with fixed interval retries.

We were able to validate this change for big ingestion (500K & 1m records) and all records are queryable in Search after this ingestion.