Add CFTC COT data integration with foundation data model layer

- New extraction package (cftc_cot): downloads yearly Disaggregated Futures ZIPs
  from CFTC, etag-based dedup, dynamic inner filename discovery, gzip normalization
- SQLMesh 3-layer architecture: raw (technical) → foundation (business model) → serving (mart)
- dim_commodity seed: conformed dimension mapping USDA ↔ CFTC codes — the commodity ontology
- fct_cot_positioning: typed, deduplicated weekly positioning facts for all commodities
- obt_cot_positioning: Coffee C mart with COT Index (26w/52w), WoW delta, OI ratios
- Analytics functions + REST API endpoints: /commodities/<code>/positioning[/latest]
- Dashboard widget: Managed Money net, COT Index card, dual-axis Chart.js chart
- 23 passing tests (10 unit + 2 SQLMesh model + existing regression suite)

Co-Authored-By: Claude Sonnet 4.6 <noreply@anthropic.com>

src/materia/pipelines.py

@@ -16,6 +16,10 @@ PIPELINES = {
         "command": ["uv", "run", "--package", "psdonline", "extract_psd"],
         "timeout_seconds": 1800,
     },
+    "extract_cot": {
+        "command": ["uv", "run", "--package", "cftc_cot", "extract_cot"],
+        "timeout_seconds": 1800,
+    },
     "transform": {
         "command": ["uv", "run", "--package", "sqlmesh_materia", "sqlmesh", "-p", "transform/sqlmesh_materia", "plan", "prod", "--no-prompts", "--auto-apply"],
         "timeout_seconds": 3600,