fix(sprint-S8.5): cross-OS CI failures — jiwer cap, Tesseract pin, %f Windows, empty inputs

feat(sprint-S8): cohérence finale — renames test dirs, /metrics endpoint, SBOM workflow

chore(sprint-S7)!: nettoyage qualité — shims supprimés, silent excepts éliminés

feat(sprint-S6)!: déploiement institutionnel — pin Tesseract, CSRF strict, logs JSON, observability

fix(sprint-S3): bugs latents — NoneType SSE, exception handler global, mypy domain strict réel

docs(sprint-S2): manifeste architectural à jour v2.0 + tests garde-fous anti-régression

docs(sprint-H.9): archive migration plans + cleanup stale doc paths

docs(sprint-H.8): cleanup obsolete legacy/shim language in production docstrings

feat(sprint-H.5+H.6)!: v2.0.0 — CLAUDE.md + CHANGELOG + bump version

feat(sprint-H.4)!: drop _legacy/ prefix — interfaces/{cli,web} consolidées

feat(sprint-H.2.c-d)!: suppression complète de adapters/legacy_engines/ et adapters/legacy_pipelines/

feat(sprint-D.2.c-f): NER, over-normalization, profile validation

feat(sprint-D.2.b): reprise sur interruption (partial_dir)

feat(sprint-H.5): cleanup baselines + tests obsolètes

feat(sprint-H.3)!: renommage reports_v2/ → reports/

feat(sprint-H.1)!: suppression cli/, llm/, extras/, fixtures.py

feat(sprint-F)!: web/ → interfaces/web/_legacy/ (Phase 9)

feat(sprint-G)!: cli/ → interfaces/cli/_legacy/ (Phase 10)

feat(sprint-E.6)!: suppression complète de picarones/measurements/

feat(sprint-E.5): 3 derniers modules measurements/ migrés — BOOTSTRAP_BASELINE = 0

feat(sprint-E.4): 3 modules hooks/registres migrés vers evaluation/metrics/

feat(sprint-E.3): metrics.py → evaluation/metrics/text_metrics.py + suppression importlib detours

feat(sprint-E.2): 10 modules measurements/ migrés vers evaluation/metrics/

feat(sprint-C)!: pipelines/ → adapters/legacy_pipelines/ (top-level supprimé)

feat(sprint-E.1): 4 métriques philological migrées vers evaluation/metrics/

feat(sprint-D.6.b)!: suppression complète de measurements/runner/

feat(sprint-D.1.d): run_benchmark_via_service — fonction publique E2E

feat(sprint-D.1.c): conversion RunResult → BenchmarkResult

feat(sprint-D.1.b): engine_to_pipeline_spec + build_adapter_resolver

feat(sprint-B): OCRLLMPipeline.run() délègue à PipelineExecutor

feat(8): importers IIIF/Gallica/eScriptorium → adapters/corpus/

feat(6-volet1): pipelines/over_normalization.py → evaluation/metrics/

test(7.D): cleanup tests architecturaux après suppression du legacy

feat(migration): Phase 7.B.3 — pipeline_benchmark consomme PipelineExecutor direct

feat(migration): Phase 7.B.2 — PipelineRunner.run délègue à PipelineExecutor

feat(migration): Lots H + I + J — statistics, htr_united/huggingface, MetricsResult

feat(migration): Lot G (partiel) — core/{diff_utils, xml_utils} supprimés

feat(migration): Lot F — picarones/report/ → reports_v2/ (suppression complète)

feat(migration): Lot E — engines/ + modules/ → adapters/legacy_*/ (suppression)

feat(migration): Lot D — measurements/{34 shims plats} → evaluation/metrics/

feat(migration): Lot C — core.{results,corpus,pipeline} → evaluation/

feat(migration): Lot B — core.{metric_registry,metric_hooks,metrics} → evaluation/

feat(migration): Lot A — core.{modules,facts} → domain/, suppression des shims

test(architecture): test parité legacy ↔ canonique + principe suppression aggressive

feat(migration): Phase 7.A.1 — engines/base.py vers evaluation/engines/base.py

test(architecture): bump BROKEN_PATHS_BASELINE to 74

feat(migration): Phase 5.E — generator + comparison + snapshot + data + templates + i18n

feat(migration): Phase 5.D — 5 vues thématiques vers reports_v2/html/views/

feat(migration): Phase 5.C batch 7 — pré-requis + 2 derniers renderers

feat(migration): Phase 5.C batch 6 — levers + philological vers reports_v2/html/