Allow DynamoDB consumers to bring their own config

[martinjlowm]

For data-rollup we heavily rely on DLQ behavior and we don't want waste retry attempts within the same Lambda invocation. Now, the consumer can provide their own config and disable retry attempts.

[martinjlowm]

meh problems with caching the value... perhaps we can do this smarter... creating a client manually for now.

[simonvandel]

In what situations do we not want to have the default retry-behaviour of the aws-sdk?
The errors they retry are transient errors, so for instance a ObjectNotFound error should not be retried by them. IMO this disabling the retry-logic will be a step back. Then spurious network errors will lead to messages in DLQ, rather than just trying the operation again which will then succeed.

[martinjlowm]

Parking this for now.