fix(review): lock OCR backend init, warn on rollback fallthrough, zero-page metric

Address PR #858 review round 2:

- OcrProcessor backend resolution is now guarded by an anyio.Lock (lazy-init,
  double-checked) so a burst of concurrent first-OCR calls resolves the backend
  once instead of each fetching its own gateway M2M token.
- The document_tier1_engine=pymupdf rollback now logs a warning when it falls
  back to the fast processor (no 'structured' registered) instead of silently
  using the very engine the operator opted out of.
- classify_from_text defaults ocr_frac to 0.0 (not 1.0) for a zero-page PDF, so
  the recorded classification metric is "fast" (no OCR evidence) rather than a
  misleading "ocr"; the no_text_layer/bad_text_layer flags are gated on having
  sampled at least one page.

New tests: zero-page classify routes fast, rollback-fallback warning.

Co-Authored-By: Claude Opus 4.8 (1M context) <noreply@anthropic.com>

tests/unit/test_doc_classifier.py

@@ -176,3 +176,12 @@ def test_classify_from_text_empty_routes_ocr():
     assert c.recommended_tier == "ocr"
     assert "no_text_layer" in c.flags
     assert c.total_chars == 0
+
+
+def test_classify_from_text_no_pages_routes_fast():
+    # An empty/corrupt PDF (no page boundaries) is not OCR evidence -> "fast",
+    # so the recorded classification metric isn't a misleading "ocr".
+    c = clf.classify_from_text("", [])
+    assert c.recommended_tier == "fast"
+    assert c.ocr_page_fraction == pytest.approx(0.0)
+    assert c.flags == set()