Refactor twin.py to improve structure and readability

[Tejas163]

No description provided.

[github-actions]

🚦 llm-twin Architectural Simulation Matrix

Automated performance validation run across target infrastructure configurations.

Profile Name	Target Model	Cluster Setup	VRAM Allocation	Throughput	Est. Latency	Status
h100_cluster_32x.yaml	LLAMA-405B	32x H100	3525.8 / 2560 GB (137.7%)	25,899.0 tok/s	41.5 ms	❌ CRITICAL OOM
h100_cluster_8x.yaml	LLAMA-70B	8x H100	216.0 / 640 GB (33.8%)	27,262.1 tok/s	14.3 ms	✅ PASSED
single_a100_40gb.yaml	LLAMA-7B	1x A100	14.8 / 40 GB (37.1%)	1,533.7 tok/s	5.7 ms	✅ PASSED
single_a100_80gb.yaml	LLAMA-34B	1x A100	74.0 / 80 GB (92.5%)	635.6 tok/s	55.7 ms	✅ PASSED
single_h100.yaml	LLAMA-70B	1x H100	159.0 / 80 GB (198.8%)	1,483.3 tok/s	114.7 ms	❌ CRITICAL OOM

Generated automatically by llm-twin continuous integration gatekeeper.