feat(web): render read tool output as a line-numbered code card

Consume the card:'read' result view (path, numbered lines, totalLines, lang)
the read backend PR added. ReadBlock (ui-primitives) draws a per-line gutter
with each line's own file number, shiki highlighting via a new highlightLines
returning per-line token arrays, a 显示 X / Y 行 window note, a height cap
matching TerminalBlock, and a copy control. read-card-model is the single
resultView derivation; a keyed ReadRow registers under read with the card
resident under its path-link summary. The generic fallback and the details
panel are read-aware. Fixture gains a windowed read turn for the built-boot
snapshot.

packages/client/ui-conversation/tests/chat-apply.spec.tsx

@@ -80,12 +80,12 @@ describe('apply wiring', () => {
     await b.runtime.dispose()
   })
 
-  it('mounts the bash sample and the todo row as keyed entries through the load-order seam', async () => {
+  it('mounts the bash sample, the read row, and the todo row as keyed entries through the load-order seam', async () => {
     const b = await bench()
-    // Both registrant plugins' inject: ['slots', 'conversation'] resolved — the
+    // All registrant plugins' inject: ['slots', 'conversation'] resolved — the
     // service being present implies the chat entry declared the hole first.
     const entries = b.slots.entries('conversation.chat.toolview')
-    expect(entries.map(e => e.options.key)).toEqual(['bash', 'todo_write'])
+    expect(entries.map(e => e.options.key)).toEqual(['bash', 'read', 'todo_write'])
     // Stats stick with the composer (not inside ChatView).
     expect(b.slots.entries('conversation.composer.dock').map(e => e.options.id)).toEqual(['stats'])
     await b.runtime.dispose()