[Autoloop: perf-comparison] Iteration 385: add SparseArray benchmark pair; fix 2 broken Python benchmarks

- Add benchmarks/tsb/bench_sparse_array.ts and benchmarks/pandas/bench_sparse_array.py
  benchmarking SparseArray.fromDense / toDense / sum / mean on 100k-element 5%-density array
- Fix escaped triple-quote docstrings in bench_str_extract_all.py and bench_str_extract_groups.py
  (re-emerged post-rebase; these caused evaluation to return null on every run)
- Metric: 726 benchmark pairs (evaluation restored from null; +1 new pair)

Run: https://github.com/githubnext/tsb/actions/runs/28663472320

Co-authored-by: Copilot <223556219+Copilot@users.noreply.github.com>

Author: github-actions[bot]

benchmarks/pandas/bench_sparse_array.py

@@ -0,0 +1,39 @@
+"""Benchmark: SparseArray fromDense / toDense / aggregations on 100k-element array (5% density)"""
+import json
+import time
+import numpy as np
+import pandas as pd
+
+N = 100_000
+WARMUP = 3
+ITERATIONS = 10
+
+# ~5% density: most values are 0, ~5k non-zero
+dense = np.zeros(N)
+for i in range(0, N, 20):
+    dense[i] = np.sin(i * 0.001) * 100 + 1
+
+# Pre-built sparse array for operations that don't test construction
+sparse = pd.arrays.SparseArray(dense, fill_value=0)
+
+# Warm up
+for _ in range(WARMUP):
+    pd.arrays.SparseArray(dense, fill_value=0)
+    sparse.to_dense()
+    sparse.sum()
+    sparse.mean()
+
+start = time.perf_counter()
+for _ in range(ITERATIONS):
+    pd.arrays.SparseArray(dense, fill_value=0)
+    sparse.to_dense()
+    sparse.sum()
+    sparse.mean()
+total = (time.perf_counter() - start) * 1000
+
+print(json.dumps({
+    "function": "sparse_array",
+    "mean_ms": total / ITERATIONS,
+    "iterations": ITERATIONS,
+    "total_ms": total,
+}))

benchmarks/pandas/bench_str_extract_all.py

@@ -1,4 +1,4 @@
-\"\"\"Benchmark: str.extractall on 10k-element string Series\"\"\"
+"""Benchmark: str.extractall on 10k-element string Series"""
 import json, time
 import pandas as pd
 

benchmarks/pandas/bench_str_extract_groups.py

@@ -1,4 +1,4 @@
-\"\"\"Benchmark: str.extract on 10k-element string Series\"\"\"
+"""Benchmark: str.extract on 10k-element string Series"""
 import json, time
 import pandas as pd
 

benchmarks/tsb/bench_sparse_array.ts

@@ -0,0 +1,43 @@
+/**
+ * Benchmark: SparseArray fromDense / toDense / aggregations on 100k-element array (5% density)
+ */
+import { SparseArray } from "../../src/index.js";
+
+const N = 100_000;
+const WARMUP = 3;
+const ITERATIONS = 10;
+
+// ~5% density: most values are 0 (fill_value), 5k are non-zero
+const dense: number[] = new Array(N).fill(0);
+for (let i = 0; i < N; i += 20) {
+  dense[i] = Math.sin(i * 0.001) * 100 + 1;
+}
+
+// Pre-built sparse array for operations that don't test construction
+const sparse = SparseArray.fromDense(dense, 0, "float64");
+
+// Warm up
+for (let i = 0; i < WARMUP; i++) {
+  SparseArray.fromDense(dense, 0, "float64");
+  sparse.toDense();
+  sparse.sum();
+  sparse.mean();
+}
+
+const start = performance.now();
+for (let i = 0; i < ITERATIONS; i++) {
+  SparseArray.fromDense(dense, 0, "float64");
+  sparse.toDense();
+  sparse.sum();
+  sparse.mean();
+}
+const total = performance.now() - start;
+
+console.log(
+  JSON.stringify({
+    function: "sparse_array",
+    mean_ms: total / ITERATIONS,
+    iterations: ITERATIONS,
+    total_ms: total,
+  }),
+);