Explain DummySpectralCNN usage in GradCAM tests

[Copilot]

Addressed review question about why tests use DummySpectralCNN instead of the real SpectralCNN model.

Response Summary

Explained the rationale for using a dummy model in unit tests:

• Test isolation: Focuses on GradCAM functionality without SpectralCNN's full complexity (BatchNorm, multiple pooling layers, embeddings)
• Test efficiency: Simpler model enables faster test execution
• Controlled scenarios: Minimal architecture simplifies verification of layer detection, edge cases, and output shapes

Identified interface mismatch: SpectralCNN returns (logits, embedding) tuple while DummySpectralCNN returns only a tensor. Offered to align the dummy model with real model behavior if desired.

---

✨ Let Copilot coding agent set things up for you — coding agent works faster and does higher quality work when set up for your repo.