An AWS Lambda function polls messages from an Amazon SQS standard queue and makes an HTTP request to an external API for each message. The company wants to ensure the Lambda function never issues more than two concurrent requests to that third-party API. Which configuration satisfies this requirement?

Setting the maximum concurrency to two on the Lambda event source mapping for the SQS queue directly limits the number of concurrent Lambda invocations processing messages from that specific SQS queue to two. This ensures that the Lambda function will never make more than two concurrent HTTP requests to the external API, as each invocation processes messages and makes a request. Provisioned concurrency guarantees a minimum number of concurrent invocations but doesn't cap the maximum. An SQS batch size of two means each invocation processes two messages, but it doesn't limit the number of concurrent invocations themselves. Lambda event filtering selects which messages to process but doesn't control concurrency.