Add AKS machine CRUD scale pipeline

[karenychen]

Summary

• Add the AKS machine CRUD perf pipeline and topology/engine steps.
• Add canadacentral scale stages for 10, 500, and 1000 nodes.
• Run 10 and 500 in parallel, then run 1000 after both complete.

Stack

• Draft PR stacked on Use az CLI for non-batch AKS machine creates #1204 so this PR shows only the non-Python pipeline/scenario diff.
• After Use az CLI for non-batch AKS machine creates #1204 merges, retarget/rebase this PR onto main before marking ready.

Scale setup

• 500 nodes uses batch API with 10 workers for a calculated batch size of 50.
• 1000 nodes uses batch API with 20 workers for a calculated batch size of 50.
• Large-scale overlay pod CIDR uses 10.128.0.0/11, aligned with existing 1000-node perf scenarios.

Validation

• yamllint -c .yamllint . --no-warnings
• focused stage dependency validation for pipelines/system/new-pipeline-test.yml
• git diff --check