docs: align benchmarking guide with harness defaults

docs/docs/extraction/benchmarking.md

@@ -68,16 +68,18 @@ active:
   test_name: null  # Auto-generated if null
 
   # API Configuration
-  api_version: v1  # v1 or v2
+  api_version: v2  # v1 or v2
   pdf_split_page_count: null  # V2 only: pages per chunk (null = default 32)
 
   # Infrastructure
   hostname: localhost
   readiness_timeout: 600
-  profiles: [retrieval]
+  compose:
+    profiles:
+      - retrieval
 
   # Runtime
-  sparse: true
+  sparse: false
   gpu_search: false
   embedding_model: auto
 
@@ -123,7 +125,7 @@ datasets:
     extract_tables: true
     extract_charts: true
     extract_images: true
-    extract_infographics: false
+    extract_infographics: true
     recall_dataset: null  # bo20 does not have recall
 
   earnings:
@@ -159,7 +161,7 @@ uv run nemo-retriever-bench --case=e2e --dataset=/custom/path
 |---------|------|--------|--------|--------|--------------|--------|
 | `bo767` | ✅ | ✅ | ✅ | ❌ | ❌ | ✅ |
 | `earnings` | ✅ | ✅ | ✅ | ❌ | ❌ | ✅ |
-| `bo20` | ✅ | ✅ | ✅ | ✅ | ❌ | ❌ |
+| `bo20` | ✅ | ✅ | ✅ | ✅ | ✅ | ❌ |
 | `financebench` | ✅ | ✅ | ✅ | ✅ | ✅ | ✅ |
 | `single` | ✅ | ✅ | ✅ | ✅ | ✅ | ❌ |