refactor: deduplicate PDF rendering by delegating to unstructured-inference (#4315)
## Summary
- Delete `_render_pdf_pages` from `pdf_image_utils.py` (~70 lines)
- Delegate `convert_pdf_to_image` and `convert_pdf_to_images` to
`unstructured-inference`'s implementation (which already has lazy
per-page rendering since v1.5.5)
- Pass `env_config.PDF_RENDER_DPI` explicitly instead of relying on
internal config
- Bump `unstructured-inference` dep to `>=1.6.2`
Peak memory for `path_only=True` drops from O(n_pages) to O(1 page) —
97% reduction on a 100-page PDF.
## Depends on
- [ ] Unstructured-IO/unstructured-inference#501 (make `dpi` explicit)
---------
Co-authored-by: codeflash-ai[bot] <178395242+codeflash-ai[bot]@users.noreply.github.com>
Co-authored-by: Kevin Turcios <turcioskevinr@gmail.com>