bugfix: Consume all result pages in DynamoDbLogicalDb.batchLoad

[szabado-faire]

This fixes a somewhat subtle bug I realized tempest2 has.

From the enhanced client docs (emphasis mine):

Partial results. A single call to DynamoDb has restraints on how much data can be retrieved. If those limits are exceeded, the call yields a partial result. This may also be the case if provisional throughput is exceeded or there is an internal DynamoDb processing failure. The operation automatically retries any unprocessed keys returned from DynamoDb in subsequent calls for pages.

Tempest currently only gets the first page of results.

From the BatchGetItem docs, the reasons that partial results can be returned are:

the response size limit is exceeded, the table's provisioned throughput is exceeded, more than 1MB per partition is requested, or an internal processing failure occurs.

The end result is that tempest silently drops some results from batchLoad when these conditions occur because it only consumes the first page of the iterator.

This might also happen in DynamoDbQueryable and DynamoDbScannable as they also only consume the first page of the iterator - I haven't done enough digging there to confirm.

Seemingly this isn't the case with tempest, as the mapper it uses handles unprocessedKeys under the hood

[kyeotic]

When I run this test locally I get an error

software.amazon.awssdk.services.dynamodb.model.DynamoDbException: Item size has exceeded the maximum allowed size (Service: DynamoDb, Status Code: 400, Request ID: 4e345a42-5d67-4d3c-82b5-34abe51e4288)
	at software.amazon.awssdk.core.internal.http.CombinedResponseHandler.handleErrorResponse(CombinedResponseHandler.java:125)

[szabado-faire]

@kyeotic not sure why a 300kB string caused problems, the max item size should be 400kB.

Dropping it to 200kB resolved that exception, and I confirmed that the test still fails if I revert the fix

[kyeotic]

Looks good! Nice job @szabado-faire