feat: Add training data transparency flags with HF Hub verification

[adityamulik]

feat: Add training data transparency flags with HF Hub verification

Problem

Currently, the AIBOM generator cannot distinguish between models where:

• Training data exists but wasn't documented
• Training data is genuinely missing or inaccessible

This creates a transparency gap where users cannot reliably audit models for complete documentation.

Reference: #80

[adityamulik]

Example of a model with available datasets

[adityamulik]

one example where dataset is linked, but the data doesn't exists