refactor(benchmark): consolidate generated artifacts under run/ directory

Restructure the benchmark pipeline to direct all simulated data, indices, statistics, and query outputs into a unified `run/` directory. Update Makefile targets, shell scripts, and Python utilities to resolve paths relative to this new base. Adjust documentation and dependency tracking to match the revised layout, and remove outdated temporary artifacts.

benchmark/compare_all_dist.py

@@ -150,17 +150,20 @@ def main() -> None:
                                  formatter_class=argparse.RawDescriptionHelpFormatter)
     ap.add_argument('--tol', type=float, default=1e-4,
                     help='Max abs diff threshold for PASS/FAIL (default 1e-4)')
-    ap.add_argument('--out', default='stats/dist_comparison/summary.csv',
+    ap.add_argument('--run-dir', default='run',
+                    help='Root directory holding reference_dist/ and obikmer_dist/ (default: run)')
+    ap.add_argument('--out', default='run/stats/dist_comparison/summary.csv',
                     help='Output summary CSV path')
     args = ap.parse_args()
 
     out_path = Path(args.out)
     out_path.parent.mkdir(parents=True, exist_ok=True)
 
+    run_dir = Path(args.run_dir)
     print(f'Comparing {len(COMPARISONS)} matrix pairs…', file=sys.stderr)
     rows = []
     for label, ref, obi in COMPARISONS:
-        rows.append(compare(label, Path(ref), Path(obi), tol=args.tol))
+        rows.append(compare(label, run_dir / ref, run_dir / obi, tol=args.tol))
 
     fields = ['comparison', 'max_abs', 'mean_abs', 'rmse', 'n_pairs', 'status']
     with out_path.open('w', newline='') as fh: