feat: Add smart paragraph merging mode with AI grouping

- Add "smart" split mode that uses LLM to intelligently merge related paragraphs
- Implement two-step approach: analyze text structure, then group by semantic relevance
- Add paragraph_merging.py with analysis and grouping prompts
- Update UI to support smart mode selection with auto-detect hint
- Add i18n translations for smart mode (en_US, zh_CN)

Co-Authored-By: Claude Opus 4.5 <noreply@anthropic.com>

pixelle_video/pipelines/standard.py

@@ -124,7 +124,13 @@ class StandardPipeline(LinearVideoPipeline):
         else:  # fixed
             self._report_progress(ctx.progress_callback, "splitting_script", 0.05)
             split_mode = ctx.params.get("split_mode", "paragraph")
-            ctx.narrations = await split_narration_script(text, split_mode=split_mode)
+            target_segments = ctx.params.get("target_segments", 8)
+            ctx.narrations = await split_narration_script(
+                text, 
+                split_mode=split_mode,
+                llm_service=self.llm if split_mode == "smart" else None,
+                target_segments=target_segments
+            )
             logger.info(f"✅ Split script into {len(ctx.narrations)} segments (mode={split_mode})")
             logger.info(f"   Note: n_scenes={n_scenes} is ignored in fixed mode")