RedshiftBatcher: on restart should consider loader offset for robustness

[alok87]

S3 batcher does shutdown gracefully but every start should consider the last uploaded offset of batcher n the loader topic last message.

This will ensure regardless of what goes wrong, the system never looses data, never reprocess same data and thus becomes Idempotent.

Idea credits: Yelp experience https://engineeringblog.yelp.com/2016/10/redshift-connector.html