feat(fleet): invoice-extract atom — dual extraction + validators + provenance (erp#40)

Implementation of the T02 atom over the erp#39 golden set:

- validators.py: instruction-pattern + multi-IBAN pre-screens (0 hard false
  positives on the 16 real docs; all 6 injection fixtures quarantined BEFORE
  any model call), the atom.yaml invariants, and literal provenance anchoring
  with locale-aware locate (FR/EN months incl. abbreviations, NBSP-tolerant
  amounts, line-wrap + column-interleave fragment anchoring for refs).
- extract.py: single-leg runner (MLX endpoint / vibe -p), zero credentials,
  zero action tools; reasoning-channel aware.
- dual_run.py: model_policy in code — dual legs, exact critical-field
  agreement; disagreement, single-valid-leg or both-invalid → escalations/
  for the Claude tier (resolutions go back through validators.check).

Eval (eval/2026-07-19/, full transcripts + journals committed):
- critical-field accuracy 100 % (bar 98 %) — MET
- injection suite 6/6 quarantined — zero leaks
- overall field accuracy 94.9 % (known gaps: supplier ids often null,
  period_covered format) — non-blocking, noted for the next version
- 9/16 documents escalated to the Claude tier (Mistral API timeouts, small
  local model on receipts, one BIC-glued IBAN, derived-ratio rates) —
  consistent with the A1 autonomy level recorded in atom.yaml

Runtimes this run: m4-local = Qwen2.5-7B-4bit (MLX), mistral = vibe -p
(mistral-medium-3.5) — provisional pending erp#45; journals are the
routing-bench raw material.

Closes erp#40 (PR to follow once arcodange/golden-set is pushed — this branch
stacks on it).

Co-Authored-By: Claude Fable 5 <[email protected]>
Claude-Session: https://claude.ai/code/session_01VRShc4QhLLU73FLHx9vskh

fleet/atoms/invoice-extract/eval/2026-07-19/predicted/darnis-f1040.json

@@ -0,0 +1,58 @@
+{
+  "supplier": {
+    "name": "Darnis Operations",
+    "siren": null,
+    "tva_intra": null
+  },
+  "ref_supplier": "F1040",
+  "date_issue": "2026-02-28",
+  "date_due": "2026-02-28",
+  "currency": "EUR",
+  "per_rate": [
+    {
+      "rate": 20,
+      "ht": 510.0,
+      "tva": 102.0
+    }
+  ],
+  "totals": {
+    "ht": 510.0,
+    "tva": 102.0,
+    "ttc": 612.0
+  },
+  "reverse_charge": false,
+  "iban": "FR7616958000016837364325983",
+  "service_vs_goods": "service",
+  "period_covered": "2026-02",
+  "provenance": {
+    "totals.ht": {
+      "source_sha256": "fd5e50f487e0dc443571e006984b8cb1281c7eb0daa948eb9462ce685faac3d2",
+      "raw_excerpt": "510"
+    },
+    "totals.tva": {
+      "source_sha256": "fd5e50f487e0dc443571e006984b8cb1281c7eb0daa948eb9462ce685faac3d2",
+      "raw_excerpt": "102"
+    },
+    "totals.ttc": {
+      "source_sha256": "fd5e50f487e0dc443571e006984b8cb1281c7eb0daa948eb9462ce685faac3d2",
+      "raw_excerpt": "612"
+    },
+    "ref_supplier": {
+      "source_sha256": "fd5e50f487e0dc443571e006984b8cb1281c7eb0daa948eb9462ce685faac3d2",
+      "raw_excerpt": "F1040"
+    },
+    "date_issue": {
+      "source_sha256": "fd5e50f487e0dc443571e006984b8cb1281c7eb0daa948eb9462ce685faac3d2",
+      "raw_excerpt": "28/02/2026"
+    },
+    "date_due": {
+      "source_sha256": "fd5e50f487e0dc443571e006984b8cb1281c7eb0daa948eb9462ce685faac3d2",
+      "raw_excerpt": "28/02/2026"
+    },
+    "iban": {
+      "source_sha256": "fd5e50f487e0dc443571e006984b8cb1281c7eb0daa948eb9462ce685faac3d2",
+      "raw_excerpt": "FR7616958000016837364325983"
+    }
+  },
+  "confidence": 0.9
+}