feat: Add workflow to generate LLM context file

[cnemri]

This PR introduces a GitHub Actions workflow that automatically concatenates README.md and all files in the docs/ directory into a single llms-full.txt file.

Motivation:

Large Language Models (LLMs) and AI agents are often trained on older versions of the MoviePy library. This llms-full.txt file provides an up-to-date, comprehensive context of the entire library's documentation, which can be used to inform and fine-tune AI agents, developers, and anyone building with the framework.

This approach is becoming a standard practice for helping AI better understand and work with modern libraries. Similar strategies are used by:

• Angular: Rules for LLMs
• Google's GenAI SDK: Codegen instructions for LLMs
• crewAI: Cursor rules in their repo
• Google ADK: Providing an llms-full.txt file

By providing this file, we make it easier for the next generation of development tools to integrate with MoviePy effectively.

[cnemri]

@Zulko @OsaAjani moviepy could be extremely helpful in the AI SWE agents era. Having this workflow could be extremely helpful. Another idea is to expose a moviepy powered MCP server to help AI Agents manipulate video, audio clips... Happy to collaborate on this.

[cnemri]

@Zulko @OsaAjani @mcorrig4 @saito828koki @keikoro would love to know your opinion on this

[saito828koki]

I basically agree with your idea. Whenever I ask Gemini to write codes using MoviePy, it writes wrong codes mostly because of the breaking changes happened in MoviePy before.

[cnemri]

I've tested the generate llm text and it provides more reliable guidance to llms like Claude and Gemini, I highly recommend adding this to repo workflow. This will update llm guidance whenever there are new updates to MoviePy. Generative Media is the next big thing and moviepy has a role to play here. @saito828koki let me know if further explanation is required on this.