When creating a claude skill for puzzletron, quantization tests fails slowing down development process

[danielkorzekwa]

I made a claude skill for puzzletron algo, then on CI a quantization test failed.

https://github.com/NVIDIA/Model-Optimizer/actions/runs/27755920751/job/82119078253?pr=1769

from pull request: #1769

it happened multiple times before, slowing down development process in modelopt.

how about making modelopt more modular so that different parts could be developed in a more isolated way?