Minimum instances, maximum instances and request concurrency express related but different requirements. A minimum cannot exceed its maximum. Concurrency must also suit the application and its downstream services; a larger number isn't automatically faster or safer.
The fictional dispatch and survey tasks supply the exact settings to save. The checker inspects those outcomes and the adapter rejects contradictory bounds. It doesn't simulate traffic, cold starts or a full provider CPU and memory compatibility matrix. A real configuration review needs measurements from the actual workload, its resource use and its dependencies. Use the local model to practice the decision and syntax, then verify production behavior independently.
Keep scale bounds consistent and distinguish settings from measurements.