feat: show model thinking/reasoning in task tree viewer

[belisarius222]

Summary

• When a sub-task's model emits chain-of-thought reasoning, the task tree viewer now shows a thinking entry at the end of the tool call list
• Selecting the thinking entry displays the full reasoning text in the detail pane (bottom third), streaming in live as CoT tokens arrive
• When reasoning text exceeds available space, the most recent tokens are shown (tail view)

Details

• New ThinkingInfo type tracks reasoning text, streaming state, and start time
• selectedSessionData memo now scans for reasoning parts alongside tool parts
• Thinking row uses ◐ icon while streaming and ● when complete, with a text preview
• ThinkingDetailView component renders in the detail section, re-rendering every tick for live updates
• Navigation (arrow keys, tab, home/end) updated to include the thinking entry in the selectable items

Test plan

• Run bun dev and trigger a task that uses a thinking model (e.g. Claude with extended thinking)
• Open task tree viewer and select the sub-task — verify "thinking" row appears at bottom of tool call list
• Select the thinking row — verify CoT text streams in the detail pane
• Verify long CoT text shows most recent content (tail behavior)
• Verify keyboard navigation (j/k, tab, home/end) works with thinking entry
• Verify non-thinking tasks still display normally

🤖 Generated with Claude Code

[github-actions]

Thanks for your contribution!

This PR doesn't have a linked issue. All PRs must reference an existing issue.

Please:

1. Open an issue describing the bug/feature (if one doesn't exist)
2. Add Fixes #<number> or Closes #<number> to this PR description

See CONTRIBUTING.md for details.