ImportError in contingency-analysis tutorial: `visualization.py` imports missing symbols (`PBELoss`, `PQ/PV/REF`)

Summary

Running the contingency tutorial notebook fails during imports because gridfm_graphkit.utils.visualization imports symbols that are not available in the current codebase.

Reproduction

1. Open examples/notebooks/Tutorial_contingency_analisys.ipynb
2. Run the imports cell containing:
   • from gridfm_graphkit.utils.visualization import (...)
3. Observe the error.

Actual behavior

Import fails with:

• ImportError: cannot import name 'PBELoss' from gridfm_graphkit.training.loss
• (after fixing that) ImportError: cannot import name 'PQ' from gridfm_graphkit.datasets.globals

Expected behavior

The tutorial import cell should run without errors in a clean install of the repository.

Root cause (likely)

visualization.py appears to use outdated imports:

• PBELoss no longer exists (or was renamed/removed) in training.loss
• bus type constants are now PQ_H, PV_H, REF_H in datasets.globals

Suggested fix

• Make PBELoss import optional (or remove hard dependency from module import path)
• Update constant imports in visualization.py to PQ_H, PV_H, REF_H (or provide backward-compatible aliases)

Environment

• OS: Linux
• Python: 3.12
• Installed from current main checkout

Start with gridfm_graphkit/utils/visualization.py and the imports cell in examples/notebooks/Tutorial_contingency_analisys.ipynb. Compare its PBELoss import with gridfm_graphkit/training/loss and its bus-type imports with gridfm_graphkit/datasets/globals, then run the notebook imports in a clean install. Done means the import cell completes without ImportError.