feat: add support for parsing and indexing plain text (.txt) files

[denis-samatov]

Summary

• Adds native support for parsing and indexing plain text (.txt) documents into PageIndex hierarchical tree structures.
• Resolves Add support for parsing .txt files #222: "Add support for parsing .txt files".
• Provides CLI integration via --txt_path in run_pageindex.py alongside --pdf_path and --md_path.

Changes

1. pageindex/page_index_txt.py:
   • extract_nodes_from_txt: extracts document structure from plain text by identifying:
     • Chapter/Section/Part/Appendix labels (e.g. Chapter 1: Intro, Section 2)
     • Numbered headings (e.g. 1. Introduction, 1.2 Details)
     • Underlined headings (=== or ---)
     • Standalone ALL-CAPS section headers
     • Graceful fallback to paragraph blocks when no explicit headings exist.
   • txt_to_tree: generates standard PageIndex structure (title, node_id, line_num, text, nodes) with support for summaries, thinning, and document descriptions.
2. pageindex/__init__.py:
   • Exposed txt_to_tree lazily in __all__ and _LAZY dictionary so top-level imports remain lightweight without eager subpackage loading.
3. run_pageindex.py:
   • Added --txt_path CLI argument.
   • Updated file argument validation to enforce specifying exactly one of --pdf_path, --md_path, or --txt_path.
   • Wired --txt_path processing and output saving to results/<doc_name>_structure.json.
4. tests/test_page_index_txt.py:
   • Added unit test suite covering numbered/chapter headers, underlined headers, all-caps headers, paragraph fallback, text content slicing, and end-to-end txt_to_tree execution.

Verification

• Ran pytest tests/test_page_index_txt.py: 6 passed.
• Ran pytest tests/test_package_surface.py: 11 passed (lazy import verification passed).
• Ran full test suite: 436 passed, 54 skipped, 0 failures.

Closes #222