Clean up affine PZ two-jet notebook diagnostics

[MoritzMaibaum]

Motivation

• Remove large, written-out polynomial trace rendering and verbose PZ multiplication/monomial diagnostics so the notebook is easier to read and maintain.
• Replace the heavy trace and rendered LaTeX output with a compact, informative summary of the final PZ two-jet components placed near the PZ norm bounds.

Description

• Updated notebooks/affine_pz_twojet_adaquad_benchmarks.ipynb to remove unused rendering imports and diagnostic helpers by deleting from IPython.display import Markdown, display and removing build_pz_twojet_norm_diagnostics, collect_pz_diagnostics, and pz_to_latex from the intervalnets imports.
• Replaced the opt-in trace path that called model.eval_pz_twojet(..., return_trace=True) and wrote a Markdown trace file with a direct call jet = model.eval_pz_twojet(cell.domain) and removed all TRACE_*, trace rendering helpers, and file I/O.
• Stopped using the diagnostic collection cell and the multiplication/monomial diagnostics, and instead constructed integrands directly from the final jet with y_sq = pz_sum_squares(jet.Y), j_sq = pz_sum_squares(jet.J), h_sq = pz_sum_squares(jet.H), and set l2_integrand, w12_integrand, and w22_integrand accordingly.
• Added a compact final diagnostics table using the summarize_final_twojet_component helper and final_twojet_monomial_diagnostics DataFrame showing distinct_monomials, max_degree, noise counts, and shapes for Y, J, and H, placed immediately before the direct PZ norm bounds cell.

Testing

• Ran automated notebook JSON assertions that validate the forbidden trace/diagnostic symbols (e.g. return_trace, TRACE_, build_pz_twojet_norm_diagnostics, collect_pz_diagnostics, pz_to_latex) are absent and that final_twojet_monomial_diagnostics and the ## Final two-jet monomial diagnostics heading are present, and all assertions passed.
• Performed a JSON load/format validation of the modified notebook, and the script checks completed successfully.

---

Codex Task