Issue 0: Cloud version: Deepgram + webhooks + Docker + JustRunMy.App

[machinatororis]

Prepare a cloud-ready version of the bot that:

• Uses the external Deepgram API for speech recognition;
• Receives updates via Telegram webhooks (FastAPI);
• Runs in a Docker container;
• Is deployed on JustRunMy.App and operates 24/7.

Subtasks:

• Add ffmpeg support to the project #1 Abstract transcription layer + configurable backends (Whisper/Deepgram)
• Implement the function convert_audio_bytes #2 Integrate Deepgram API as external speech-to-text backend
• Refactor the project structure #3 Add FastAPI webhook server for Telegram updates
• Create the module utils/audio.py #4 Dockerize the project (image for cloud deployment)
• Create the handlers/voice.py module #5 Deploy to JustRunMy.App and document deployment
• Create main.py as the application entry point #6 Translate comments/docstrings to English (polish for portfolio)