Split long text and Markdown files into sized chunks with token estimation and sliding overlap windows for RAG and vector databases.