Add cloud-based Whisper support for transcription

[psmoros]

This PR adds support for OpenAI's cloud-based Whisper API as an alternative to the local Whisper model. This enhancement provides several benefits:

Key Features:

• Cloud-based Whisper integration via OpenAI's API
• Accurate word-level timestamp generation for better animation synchronization
• Compatible with ARM64 architectures (e.g., Apple Silicon)
• No need to install large local Whisper models

Implementation Details:

• Added use_cloud_whisper flag to enable cloud-based transcription
• Created custom entry point script (manim_cloud_whisper.py) for easy usage
• Updated SpeechService class to handle cloud-based transcription
• Added environment variable support for configuration
• Improved word boundary processing for accurate timing

Usage:
Users can enable cloud-based Whisper in three ways:

1. Using the custom entry point: python manim_cloud_whisper.py
2. Setting environment variable: MANIM_VOICEOVER_USE_CLOUD_WHISPER=1
3. Programmatically: use_cloud_whisper=True in OpenAIService

Example:
Added a comprehensive linear regression demo showcasing the cloud-based Whisper functionality with synchronized voiceovers and animations.

Note: Requires OpenAI API key to be set in environment variables.