[Feature] Make Istio control plane resource requests configurable

**Is your feature request related to a problem? Please describe.**
We use the Istio service mesh add-on for AKS. In clusters under high load (~1400 services, ~2000 pods in our application namespace), the istiod control plane pods consume significantly more memory and CPU than their resource requests — up to 7.7 Gi memory (requested: 2Gi, ~3.8× over) and 6 CPU cores (requested: 500m, ~12× over).

Because the scheduler uses requested values for bin-packing decisions, this massive under-request causes memory pressure on system nodes, leading to frequent evictions of other pods. Since istiod resource requests are managed by AKS and not configurable by the user, we have no way to mitigate this.

**Describe the solution you'd like**
It would be great if there was a way to configure requested memory and CPU for Istio control plane pods.

**Describe alternatives you've considered**
I tried increasing the number of system nodes to match the number of istiod pods so that each istiod pod can be placed on a separate system node. It helped to some extent but it didn't resolve the issue completely.
I specified `discoverySelectors` to limit how many namespaces Istio has to monitor.

**Additional context**
Memory usage of istiod discovery container

CPU usage of istiod discovery container