lotus bot - agent tutor

**Description**

One of lotus bot's key features is to be an AI tutor, one that is designed to build on the user's knowledge and understanding on a topic provided but not give a direct answer immediately, walk the user through their current problem, take the steps to help them, determine whether or not they would need more guidance on a particular topic, and provide answers only when necessary / checking the answer.

This will be implemented as a sub agent of the current agent.

The user can dynamically switch between being primarily a tutor bot, or a regular chatbot.
(In beta, it would just be swapped using a specific keyword)

**Features**

- The user can enter a custom URL to a review document, Quizlet, or study guide:
This data will be embedded into a VDB, chunked for context, queried by the AI, and used by the Agent to guide the user.
This will be stored in Fields: Context, Question, Answer
- The AI could use sentiment analysis to determine the understanding of the user.
- The AI could also ask the user to explain their understanding of the question.
This also may be scaled based on question understanding, for example, if a question is correct immediately, then this prompting could happen
- The user could also ask the bot to give sample questions and understanding checkpoints.

**Reasoning**

- To provide an excellent support resource without having the AI overlap understanding on a topic
To help other students (as well as myself) be able to use AI as an exceptional learning tool
- Have an AI have direct access to review resources for in-depth analysis and review

The issue describes a new sub-agent for tutoring. Start by examining the existing agent structure in the repository to understand how sub-agents are implemented. Look for code related to vector databases (VDB), chunking, and embedding. The goal is to design and integrate a system that can switch modes, process custom URLs, and perform sentiment analysis. 'Done' means a functional tutor agent that can be toggled and uses provided study materials.

Escrita pelo modelo de indexação a partir do texto da issue.