fix(ocr): emit valid Markdown newlines for PPTX conversion

[Leonardo Montemurro (MontesanoDev)]

Closes #2010

Summary

PptxConverterWithOCR emitted literal \n sequences instead of actual line breaks in Markdown output.

The optional LLM caption path also imported llm_caption from a module that does not exist in the markitdown_ocr package.

Changes

• Replace literal \n sequences with actual newlines in PPTX Markdown output.
• Import llm_caption from markitdown.converters._llm_caption.
• Update PPTX OCR snapshots to expect valid Markdown line breaks.
• Add regression tests for newline output and the LLM caption import path.

Testing

pytest packages/markitdown-ocr/tests/test_pptx_converter.py -v

Result: 8 passed.

[Leonardo Montemurro (MontesanoDev)]

@microsoft-github-policy-service agree

[Adam Fourney (afourney)]

Thank you for contributing the newline and caption-import fixes, including the regression tests.

#2506, included in v0.1.8, replaced the duplicated PPTX OCR conversion paths with the core presentation pipeline. This addresses both the literal backslash-n formatting and the incorrect caption-module import; the current formatting and caption regressions pass.

Closing this PR as superseded. Please use markitdown==0.1.8 with markitdown-ocr==0.1.1. Please reopen this PR if it still addresses a case that the current implementation misses, with a reproduction or a description of the remaining difference.