cli: parallelize formatter calls with thread pool

reformat_draft now uses concurrent.futures.ThreadPoolExecutor with a
default of 8 workers. OpenAI SDK calls are synchronous but network-
bound, so threads release the GIL during I/O and give real speedup.
ThreadPoolExecutor.map preserves input order regardless of completion
order, so output is deterministic.

Empirical: HML draft (134 entries) went from ~25 min serial to 1m55s
with concurrency=12. ~12x speedup; further increases hit OpenAI rate
limits.

CLI gains a --concurrency flag (default 8) for tuning per draft size /
rate limit headroom. concurrency=1 forces serial execution for
debugging. New test asserts that order is preserved when concurrent
calls finish out of input order (uses a sleep-by-index fake formatter).

tests/test_cli.py

@@ -66,3 +66,38 @@ def test_reformat_draft_bibliography_heading_preserved():
     draft = "## Bibliography\n\nentry.\n"
     output = reformat_draft(draft, formatter=_fake_formatter)
     assert "## Bibliography" in output
+
+
+def test_reformat_draft_preserves_order_under_concurrency():
+    # With concurrent execution the formatter is called on all entries in
+    # parallel; the CLI must reassemble them in input order regardless of
+    # which call finishes first. This test uses a fake formatter that
+    # sleeps based on the entry content so earlier entries finish AFTER
+    # later ones if order were naively tied to completion.
+    import time
+
+    def slow_fake(messy: str) -> str:
+        # Entries with lower index sleep longer so they finish last.
+        n = int(messy.split()[-1])
+        time.sleep(0.05 * (5 - n))
+        return f"FORMATTED({messy})"
+
+    draft = """\
+## Bibliography
+
+entry 0
+entry 1
+entry 2
+entry 3
+entry 4
+"""
+    output = reformat_draft(draft, formatter=slow_fake, concurrency=4)
+    # Entries must appear in input order.
+    lines = [l for l in output.splitlines() if l.startswith("FORMATTED(")]
+    assert lines == [
+        "FORMATTED(entry 0)",
+        "FORMATTED(entry 1)",
+        "FORMATTED(entry 2)",
+        "FORMATTED(entry 3)",
+        "FORMATTED(entry 4)",
+    ]