bench(valid): a training freeze may cap the history walk at the newest N first-parent commits

ordinary_history(limit=) walks only the newest commits; assemble.freeze(history_limit=) and
--history-limit pass it through and the record carries it. The held-out benchmark passes none.
Reason: trigger.dev's 7,109-commit walk runs for hours at the per-commit cost of the function
and generated checks; the training freeze now walks its newest 1,500.

Co-Authored-By: Claude Fable 5.1 <[email protected]>

tests/test_unseen_extract.py

@@ -670,3 +670,12 @@ def test_partial_clone_lazy_fetch_preserves_extraction_content(history, tmp_path
     assert partial.files_read > 0 and partial.bytes_read > 0
     assert json.dumps(result, sort_keys=True).encode() == json.dumps(baseline, sort_keys=True).encode()
     assert sum("built-in: git fetch " in line for line in trace.read_text().splitlines()) >= 12
+
+
+def test_history_limit_walks_only_the_newest_commits(history):
+    git, *_ = history
+    full = e.ordinary_history(git, "HEAD")
+    assert len(full) > 2
+    assert e.ordinary_history(git, "HEAD", limit=2) == full[:2]  # the newest first-parent commits, same rows
+    with pytest.raises(ValueError, match="positive integer"):
+        e.ordinary_history(git, "HEAD", limit=0)

tests/test_valid_benchmark.py

@@ -448,7 +448,7 @@ def test_insufficient_negatives_and_invalid_ratio(pair, tmp_path, monkeypatch):
     with pytest.raises(ValueError, match="ordinary_floor"):  # Rejected is a ValueError with a fixed reason code
         assemble.freeze(tmp_path / "default", **kwargs)
     result = assemble.freeze(tmp_path / "train", floor=20, **kwargs)
-    assert result["ordinary_floor"] == 20 and result["negative_ratio"] == 20
+    assert result["ordinary_floor"] == 20 and result["negative_ratio"] == 20 and result["history_limit"] is None
     # Twenty eligible commits for a ratio of twenty: every one of them is drawn, none twice.
     ordinary = [c for c in replay.load_slice(tmp_path / "train/pool-valid.toml", 1) if c["kind"] == "ordinary"]
     assert len(ordinary) == 20 and len({c["head"] for c in ordinary}) == 20