docs: add UI vs text chatbot performance article

[nicovaleops]

Closes #9

Summary

• Adds a research-backed article comparing UI-based chatbots and text-based chatbots across task completion rate, time-on-task, and user satisfaction.
• Positions OpenUI as the practical developer path from plain generated text to structured, interactive task output.
• Includes a measurement model teams can use for A/B testing UI-based assistant outputs against text-only outputs.

Validation

• Word count: 1,509 words.
• Private/internal term scan passed.
• git diff --check passed.