Get docling fixes onto main

[jkwatson]

No description provided.

[Copilot]

Pull Request Overview

This PR updates the docling integration by bumping its version and replacing the existing hierarchical chunker with a new hybrid chunker implementation in the document reader.

• Bump docling dependency to >=2.40.0 and add docling-ibm-models override
• Swap HierarchicalChunker for HybridChunker and remove manual serialization
• Clean up related imports

Reviewed Changes

Copilot reviewed 2 out of 3 changed files in this pull request and generated no comments.

File	Description
llm-service/pyproject.toml	Upgraded docling version and added docling-ibm-models override
llm-service/app/ai/indexing/readers/docling_reader.py	Switched to HybridChunker, removed per-item serialization, updated imports

Comments suppressed due to low confidence (5)

llm-service/app/ai/indexing/readers/docling_reader.py:45

• Duplicate import of BaseChunk detected on line 47 as well. Remove one of these to avoid confusion.

from docling_core.transforms.chunker import BaseChunk

llm-service/app/ai/indexing/readers/docling_reader.py:49

• SerializationResult is no longer used after removing manual serialization. Consider removing this import.

from docling_core.transforms.serializer.base import SerializationResult

llm-service/app/ai/indexing/readers/docling_reader.py:50

• MarkdownDocSerializer import is unused since serialization was removed. You can delete this import.

from docling_core.transforms.serializer.markdown import MarkdownDocSerializer

llm-service/app/ai/indexing/readers/docling_reader.py:72

• MarkdownSerializerProvider is not imported, so this will raise a NameError. Add from docling_core.transforms.serializer.markdown import MarkdownSerializerProvider (or correct provider) to the imports.

        chunky_chunks = HybridChunker(serializer_provider=MarkdownSerializerProvider()).chunk(docling_doc.document)

llm-service/app/ai/indexing/readers/docling_reader.py:84

• Variable document is not defined in this scope. Did you mean to use docling_doc.document.metadata or pull metadata from docling_doc?

            node.metadata["file_name"] = document.metadata["file_name"]