AL2023 should use actual max pods to calculate kube-reserved

[stevehipwell]

What would you like to be added:

I'd like to see AL2023 (AL2 doesn't have the correct inputs) calculate the kube-reserved value from the actual max pods if provided and not always use the hypothetical instance value.

Why is this needed:

Running a calculation for kube-reserved based on max pods makes sense but using the wrong value for max pods in this calculation when there is a user defined value is part of the structured API doesn't make any sense. If I've set my node's max pods value to 110 then I expect the kube-reserved value to reflect this.

CC @cartermckinnon