Return partial batch failures from an SQS-triggered Lambda function using ReportBatchItemFailures instead of failing the whole batch

Steps

1. Update the event source mapping to include ReportBatchItemFailures in FunctionResponseTypes, e.g. `aws lambda update-event-source-mapping --uuid <uuid> --function-response-types ReportBatchItemFailures`
2. In the function, process each record and collect the message IDs that failed
3. Return {"batchItemFailures": [{"itemIdentifier": "<message-id>"}, ...]} — an empty list, null, or empty response means the whole batch succeeded
4. For a standard queue, only the listed failed message IDs become visible again for redelivery; successfully processed ones are not retried
5. For a FIFO source queue, stop processing at the first failure and report that message plus every subsequent unprocessed message as failed, to preserve ordering
6. Validate the response shape carefully — malformed JSON, an empty-string or null itemIdentifier, a bad key name, or an ID not present in the batch all cause AWS to treat the entire batch as failed