Improve receive responses from the bot to almost instantly

[maxzaikin]

Description

The current RAG pipeline has a high response latency (3-5 minutes) due to the slow inference speed of the local LLM. This leads to client timeouts and a poor user experience. This story aims to solve the UX problem by implementing streaming, allowing the user to see the response as it's being generated.

Acceptance Criteria

• The API has a new /chat/stream endpoint that returns a StreamingResponse.
• The RAGEngine can generate and yield tokens in real-time.
• The tg-gateway client can consume the stream and progressively edit the message in Telegram.
• The user sees the first token of the response within 2-3 seconds of sending a message.

Tasks

• [Task] ID: RAG-10 - feat(api): Refactor the RAG API endpoint to support streaming responses.
• [Task] ID: RAG-11 - feat(agent): Modify RAGEngine to return an async generator.
• [Task] ID: RAG-12 - refactor(tg-gateway): Update RagApiClient to handle streaming HTTP responses.
• [Task] ID: RAG-13 - feat(tg-gateway): Implement a message handler that edits a Telegram message progressively.