Add PZ two-jet norm diagnostics

[MoritzMaibaum]

Motivation

• Provide a small diagnostic API so callers can inspect the final polynomial-zonotope two-jet before the norm integrand construction and subsequent integration steps.
• Make it easy to collect structural metadata and rendered snippets for debugging term growth and approximation noise without changing propagation semantics.

Description

• Add build_pz_twojet_norm_diagnostics(jet: PZTwoJet, *, include_integrands=True, render=True, max_terms=12, precision=4) in src/intervalnets/pz_norms.py that returns raw Y, J, H references, metadata summaries, optional LaTeX/Markdown renderings, and optional cached l2_integrand, w12_integrand, and w22_integrand objects.
• Introduce _pz_metadata_summary to compute term counts, max degree, shapes, num_noise, and noise-kind counts using collections.Counter.
• Export build_pz_twojet_norm_diagnostics from the package public API in src/intervalnets/__init__.py.
• Add tests in tests/test_pz_norms.py asserting the diagnostics helper does not mutate the input PZTwoJet and that the produced integrands match the existing pz_twojet_l2_integrand, pz_twojet_w12_integrand, and pz_twojet_w22_integrand helpers, and update the notebook notebooks/affine_pz_twojet_adaquad_benchmarks.ipynb to build diagnostics before integrating cached integrands.

Testing

• Ran the unit tests with python -m pytest tests/test_pz_norms.py -q and all tests passed (17 passed).
• Verified syntax/byte-compile with python -m compileall -q src/intervalnets tests/test_pz_norms.py which completed successfully.

---

Codex Task