fix: accept string paths in corpus extraction

[gaoflow]

Summary

• normalize string file paths to Path at the extract_corpus_parallel boundary before slicing and token packing
• keep internal extraction unit annotations aligned with existing Path | FileSlice behavior
• add a regression test for list[str] input

Fixes #1386

Testing

• uv run pytest tests/test_chunking.py -q
• uv run pytest tests/ -q
• uv run ruff check graphify/llm.py tests/test_chunking.py
• git diff --check
• uv run graphify update .

AI assistance was used under my direction.