fix(terminal): use grapheme_width for accurate column slicing

text_in_cols and text_slice_cols now walk grapheme clusters with
grapheme_width() instead of assuming 1 char = 1 column.  This fixes
column slicing for:
- Wide characters (CJK, emoji) that occupy 2 grid columns
- Combining marks that occupy 0 columns

The prior 1:1 char∶col mapping caused selection-copy and serialize
output to misalign column ranges when wide/combining characters were
present in a row.

- Update roadmap: mark both P0 follow-up items as resolved

TERMINAL_ROADMAP.md

@@ -62,7 +62,7 @@
    - 实现于 `component.rs`（`Selection`/`selection_rects`/`point_to_cell`/鼠标处理）+ `core.rs::selection_text` + `core.rs::TerminalRow::text_in_cols`，均有单元测试。
 3. **快捷键**：Ctrl-Shift-C 复制 / Ctrl-V 粘贴。
 
-**已知后续项（非阻塞）**：拖拽到视窗边缘时**没有自动滚动延伸选区**（需先滚到位再选）；行内 wide+组合字符混排的列切片为近似。
+**已知后续项（非阻塞）**：拖拽到视窗边缘时**没有自动滚动延伸选区**（✅ 已通过上一轮 auto-scroll 实现）；行内 wide+组合字符混排的列切片（✅ 已通过 grapheme_width 修正 text_in_cols / text_slice_cols）。
 
 ---