Include PPTX Speaker Notes in HybridChunker Output

### Requested feature
Currently, speaker notes are successfully extracted from PPTX files after [PR #474](https://github.com/DS4SD/docling/pull/474), using the export_to_markdown() method with the appropriate ContentLayer configuration (e.g., BODY and FURNITURE).

However, when using HybridChunker, the resulting chunks do not include the speaker notes, even when those notes are present in the parsed content.

### Requested enhancement:
Update HybridChunker to include speaker notes in the final chunk output when they are available. This would ensure that all relevant slide content including what's often found in the speaker notes (context, talking points, or clarifications) is preserved in downstream processing.

This is especially useful for LLM-based pipelines like RAG, where context-rich inputs improve the quality of retrieval and responses.

Start by tracing HybridChunker output for PPTX content and compare it with export_to_markdown() using the BODY and FURNITURE ContentLayer configuration. Confirm that available speaker notes are included in the resulting chunks and remain available for downstream RAG processing.

Written by the indexing model from the issue text.