Advanced AI Prompt for Lambda Performance Tuning with CloudWatch

Act as an AWS Serverless Performance Engineer. Perform a deep analysis of the native CloudWatch metrics emitted by the high-throughput [FUNCTION_NAME] Lambda function. Architect a suite of highly sensitive CloudWatch Alarms: 1) Monitor `MemoryUtilization` against the [MEMORY_UTILIZATION_TARGET] to detect impending Out-Of-Memory (OOM) crashes. 2) Track the `InitDuration` metric, triggering an alert if cold starts exceed the [COLD_START_THRESHOLD]. 3) Monitor `Throttles` based on the strict [CONCURRENCY_LIMIT] quota. Provide a sophisticated Python/Boto3 script that programmatically queries the CloudWatch API to extract these historical data points and outputs a mathematical recommendation for the exact memory allocation (MB) required to hit the absolute lowest cost-duration (GB-seconds) execution 'sweet spot'.