added regression tests for official models - fix for #164

[nkundiushuti]

• Regression test: Slow integration test fingerprints official HF models on a deterministic labeled batch so loaded weights match expected outputs.
• Maintenance: Script regenerates those fingerprints when checkpoints change on purpose.
  Fix for Comprehensive integration test on labelled data required to confirm model checkpoints load correctly #164

[Copilot]

Pull request overview

Adds a slow integration regression test to ensure “official” ESP Hugging Face checkpoints produce stable numerical outputs on a deterministic mini-batch, plus a helper script to regenerate the expected output fingerprints when checkpoints intentionally change (addresses #164).

Changes:

• Introduces a slow integration test that loads each official HF-backed ESP model and asserts a SHA-256 fingerprint of pooled outputs matches a hardcoded reference table.
• Adds a regeneration script to recompute and print updated fingerprints (with optional JSON output).

Reviewed changes

Copilot reviewed 2 out of 2 changed files in this pull request and generated 5 comments.

File	Description
tests/integration/test_official_models_output_regression.py	New slow integration test that fingerprints pooled model outputs for all official HF-backed esp_ models and enforces reference coverage.
scripts/regenerate_official_model_output_fingerprints.py	Utility to regenerate the fingerprint mapping used by the integration regression test.

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.

[GaganNarula]

I think I agree with all the copilot comments. Other than that it looks good