Store records of everything #8
Description
Given the logic in this service is related to billing and accounting, it's interesting to keep copies of every document involved in case something goes wrong and we need to do some manual investigation.
One main thing we are not storing right now are receipt batches as created by the nodes. We fetch and process them, but we don't store our own copies. We could store them in a bucket with a retention period that matches the most restrictive regulations about expiration of administrative issues.
Other interesting pieces of data, such as the original egress/track invocation, or the egress/consolidate invocations and receipts, are already being stored in DynamoDB tables (we may consider moving these to S3 buckets as a cost optimization measure).
Refs: #7 (comment), #7 (comment)