[Feature] API-first managed scheduling

**Is your feature request related to a problem? Please describe.**

Running batch, HPC, or AI workloads on AKS requires deploying and managing a custom job/pod admission logic, custom scheduler (e.g.,Volcano) and integrating it with autoscaling and placement logic.

This creates significant complexity, operational overhead

**Describe the solution you'd like**
A managed, Kubernetes-native batch scheduling capability in AKS that:
1. Provides job queueing and admission control
2. Integrates with cluster autoscaling (CAS / NAP)
3. Supports constraint-aware placement (e.g., resources, topology, external signals)
4. Works with standard Kubernetes APIs (Jobs, JobSets, etc.)
5. Removes the need to operate a separate scheduler

**Expected behavior**
Customers explicitly require a “behind an API” experience, where:
User submits workload → Scheduling decisions handled by platform → Placement constraints enforced → Autoscaling + lifecycle managed automatically

Alternatives you've considered
1. Deploying OSS schedulers (Volcano, Slurm integrations) and job queues (Kueue)
2. Building custom controllers / admission webhooks