Strip JSX / HTML tags from llms-full file

[lukecd]

It would be useful to have an option to generate LLMs-full with JSX and other HTML tags removed.

Why this helps:

• JSX/HTML adds token bloat with little value for LLM consumption
• Increases context size without improving semantic signal
• Plain text improves chunking, embedding quality, and prompt efficiency

Suggested behavior:

• Preserve headings and structure, but emit clean markdown or plain text only

Happy to help test or provide examples if useful.