upgrade: adopt V8.31.1 (theme 1.3/1.4 redesigns, editor deep-canvas, guided launch, component library)

Overlay 190 upstream items (V8.22→8.31) preserving trade-pro theme,
registry tests, and design docs. Re-applied full patch set on identical
anchors (discovery/env_for/blocks.css/type-mapping/locale-override/
theme_color/frontend entries/test_theme_pro_v3 mods) plus compose pins.

Verified: theme suites 35/35; dual-baseline full run 415 passed vs
pristine 386+29 ours, FAILED-set diff EMPTY (same 10 Windows env fails).
Live: container 8.31.1-dev.1 healthy, hesupipe 192 entries intact,
trade-pro build 203 pages 0 errors.

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

scripts/regression_v821.py

@@ -84,7 +84,7 @@ def main(argv=None):
             if result['status']!='PASS':break
         if results[-1]['status']!='PASS':break
     success=len(results)==args.repeat*((len(modules)+args.batch_size-1)//args.batch_size) and all(x['status']=='PASS' for x in results)
-    info={'version':'8.21.3-dev.1','generated_at':datetime.datetime.now(datetime.timezone.utc).isoformat(),
+    info={'version':'8.25.0-dev.1','generated_at':datetime.datetime.now(datetime.timezone.utc).isoformat(),
           'status':'PASS' if success else 'FAIL','test_modules':len(modules),'repeats':args.repeat,
           'tests_per_round':sum(x['passed'] for x in results if x['name'].startswith('round1-')),
           'total_passes':sum(x['passed'] for x in results),'batches':results,