Scaffold CMOS 18 reformatter: harness, linter, parser, formatter, CLI

Initial phase-1 baseline of the karpathy/autoresearch-style loop.
The formatter module is the inner-loop artifact; parser and linter
are infra. The linter carries a LINTER_VERSION hash (v0.2.0) that
will force a re-baseline on any rule change.

Components:
- harness/diff.py: case-sensitive field-level substring diff
- harness/score.py: three-axis scoring (field, linter, canary exact)
- src/cmos/linter.py: 9 CMOS 18 structural rules, each Purdue/CMOS cited
- src/cmos/parser.py: locate ## Bibliography section, split entries
- src/cmos/formatter.py: prompt + OpenAI call with caller injection
- src/cmos/cli.py: cmos format path/to/draft.md
- scripts/run_loop.py: loop runner with --fake mode for no-API runs
- exemplars/: 3 canary seed exemplars (book, journal w/DOI, web),
  sourced from chicagomanualofstyle.org quick guide

Tests: 48 passing. Fake-mode baseline scalar = 0.000 on the 3 seed
exemplars (identity caller fails the linter on every rule). This is
the floor the real GPT-5 formatter needs to improve from.

tests/test_cli.py

@@ -0,0 +1,68 @@
+"""Tests for src/cmos/cli.py — end-to-end draft reformatting.
+
+The CLI orchestrates parser → formatter → reassemble. Tests inject a fake
+formatter so no API calls happen; the goal is to pin the glue logic, not the
+LLM behavior.
+"""
+
+from pathlib import Path
+
+import pytest
+
+from cmos.cli import reformat_draft
+from cmos.parser import NoBibliographyError
+
+
+def _fake_formatter(messy: str) -> str:
+    # Deterministic fake: return a cleaned-up marker string so the output is
+    # recognizable.
+    return f"FORMATTED({messy.strip()})"
+
+
+def test_reformat_draft_rewrites_only_bibliography_section():
+    draft = """\
+# My Paper
+
+Some prose with citations.
+
+## Bibliography
+
+yu, charles. interior chinatown. 2020.
+kwon, hyeyoung. inclusion work. 2022.
+"""
+    output = reformat_draft(draft, formatter=_fake_formatter)
+    assert "# My Paper" in output
+    assert "Some prose with citations." in output
+    assert "FORMATTED(yu, charles. interior chinatown. 2020.)" in output
+    assert "FORMATTED(kwon, hyeyoung. inclusion work. 2022.)" in output
+    # Original messy lines must not also survive.
+    assert "yu, charles. interior chinatown. 2020." not in output.replace(
+        "FORMATTED(yu, charles. interior chinatown. 2020.)", ""
+    )
+
+
+def test_reformat_draft_preserves_sections_after_bibliography():
+    draft = """\
+## Bibliography
+
+one entry.
+
+## Appendix
+
+appendix text.
+"""
+    output = reformat_draft(draft, formatter=_fake_formatter)
+    assert "FORMATTED(one entry.)" in output
+    assert "## Appendix" in output
+    assert "appendix text." in output
+
+
+def test_reformat_draft_raises_when_no_bibliography():
+    with pytest.raises(NoBibliographyError):
+        reformat_draft("## Intro\n\nno bib here.\n", formatter=_fake_formatter)
+
+
+def test_reformat_draft_bibliography_heading_preserved():
+    draft = "## Bibliography\n\nentry.\n"
+    output = reformat_draft(draft, formatter=_fake_formatter)
+    assert "## Bibliography" in output