Restructure: descriptive tier and experiment names, paper/manuscript

- paper/pnas -> paper/manuscript (venue-neutral)
- configs/layer1 -> configs/inheritance, src/knowledge -> src/inheritance
  (imported as `inheritance`), make layer1 -> make inheritance; layer2 alias dropped
- inheritance and trained-network bundles named after the manuscript figure
  they feed (fig2_grounding_sweep, figS3_rebaselining, ...), or descriptively
  where they feed none; configs keep their `experiment:` value so parquet
  hashes are unchanged, only output.dir moves
- figure scripts, SI figure sources, notebooks, REPRODUCING.md, README and the
  SI Methods/tables updated; make clean no longer deletes tracked manifests;
  reproduce.sh hashes the s{seed}/ layouts too

Co-Authored-By: Claude Fable 5.1 <noreply@anthropic.com>
Claude-Session: https://claude.ai/code/session_01Y64o8FKP7rCuXzC48pxpMm

figures/plot_E8.py

@@ -1,67 +0,0 @@
-"""E8 figure — the vertical claim: n-parent recombination exceeds any parent (Fisher–Muller).
-
-The society headline. Decorrelated *parents* are specialists (expert on some loci, agnostic on the
-rest); an *offspring* recombined from all of them can be fitter than any parent — capability that
-*exceeds* every component, not just recovers a ceiling. Unlike biological sex there is no two-parent
-limit, so capability climbs toward the optimum as the parent pool grows and decorrelates.
-
-Two panels: (A) deployed capability (mode-genotype fitness) vs parent count at ρ=0 — sexual
-recombination reaches the optimum (a genotype no parent had) while the best single parent and the
-mean-mixture "model soup" plateau below; (B) the decorrelation control — sexual capability vs parent
-count for ρ ∈ {0, 0.5, 1}: decorrelated parents (ρ=0) climb to the optimum, identical parents (ρ=1)
-buy nothing. Reads only the committed bundle.
-
-Usage:  python figures/plot_E8.py [results/E8]
-"""
-
-from __future__ import annotations
-
-import sys
-from pathlib import Path
-
-import matplotlib.pyplot as plt
-import numpy as np
-
-sys.path.insert(0, str(Path(__file__).parent))
-from _figlib import load_bundle, mean_ci, savefig, letter_axes  # noqa: E402
-
-
-def main(results_dir: str = "results/E8") -> None:
-    df, cfg = load_bundle(results_dir)
-    L = cfg["society"]["L"]
-    rhos = sorted(df["rho"].unique())
-
-    fig, axes = plt.subplots(1, 2, figsize=(13, 5))
-
-    # Panel A: best-parent vs average vs sexual, at rho=0.
-    ax = axes[0]
-    d0 = df[df["rho"] == 0.0]
-    for col, c, lab in [("best_parent", "#7f7f7f", "best single parent"),
-                        ("average", "#1f77b4", "average (model soup)"),
-                        ("sexual", "#d62728", "sexual recombination")]:
-        k, m, ci = mean_ci(d0, "K_T", col)
-        ax.errorbar(k, m, yerr=ci, fmt="-o", color=c, capsize=3, label=lab)
-    ax.axhline(L, ls=":", color="green", lw=1, label=f"optimum ($L$={L})")
-    ax.set(xlabel="number of parents $K_T$", ylabel="deployed capability (mode fitness)",
-           title="Recombination exceeds any parent (ρ=0):\nsexual reaches the optimum; soup & best-parent plateau")
-    ax.legend(frameon=False, fontsize=9)
-
-    # Panel B: sexual capability vs K_T for each rho (decorrelation control).
-    ax = axes[1]
-    colors = plt.cm.viridis(np.linspace(0, 0.8, len(rhos)))
-    for rho, c in zip(rhos, colors):
-        sub = df[df["rho"] == rho]
-        k, m, ci = mean_ci(sub, "K_T", "sexual")
-        ax.errorbar(k, m, yerr=ci, fmt="-o", color=c, capsize=3, label=fr"ρ={rho:g}")
-    ax.axhline(L, ls=":", color="green", lw=1, label=f"optimum ($L$={L})")
-    ax.set(xlabel="number of parents $K_T$", ylabel="sexual-recombination capability",
-           title="Decorrelation is the fuel:\nρ=0 climbs to the optimum; ρ=1 (clones) buy nothing")
-    ax.legend(frameon=False, fontsize=9)
-
-    fig.tight_layout()
-    letter_axes(fig)
-    savefig(fig, results_dir, "E8")
-
-
-if __name__ == "__main__":
-    main(*sys.argv[1:])