Add weight-trap histogram plotting helpers and integrate into WW checkpoint plots

[charlesmartin14]

Motivation

• Provide visual diagnostics for dominant weight values ("traps") in converted weight matrices alongside existing WeightWatcher outputs to aid analysis of training checkpoints.

Description

• Import matplotlib.pyplot and add _extract_weight_matrix to obtain 2D weight arrays from layer objects.
• Add _dominant_weight_values to pick top unique weight magnitudes and plot_trap_lines_on_weight_histogram to draw trap vertical lines and labels on a weight histogram, plus wrapper plot_removed_traps_on_weight_histogram.
• Introduce TRAP_HIST_MAX_LINES and call plot_trap_lines_on_weight_histogram for W2, W7, and W8 when make_ww_plots is enabled, saving figures to the round plot directory.

Testing

• No automated tests were added or executed for this change.

---

Codex Task