docs(extraction): sync post-#2179 extraction doc fixes to main

[kheiss-uwzoo]

Summary

Sync extraction doc structure on main with post-#2179 review fixes that landed on 26.05 in PR #2194 but never reached main. This is not a literal cherry-pick of #2194 — review feedback on this PR evolved the approach.

What changed

• Remove custom-metadata.md — consolidate metadata/filtering guidance into vdbs.md#metadata-and-filtering; add mkdocs.yml redirect; update cross-links in workflow-agentic-retrieval.md, integrations-langchain-llamaindex-haystack.md, and the doc-snippet test list.
• Fix stale caption anchors — retarget FAQ and multimodal-extraction.md links from prerequisites-support-matrix.md#image-captioning-2605 to in-page anchors; rename the support-matrix section to #image-captioning-nim-hardware with a legacy #image-captioning-2605 span for external bookmarks; fix deployment-options.md to use the new anchor.
• Trim support-matrix admonition — remove the chart-caption admonition block per review; add one sentence of caption-scope prose under #image-captioning in multimodal-extraction.md.
• Other fix audio-video.md markdown rendering (follow-up to #2179) #2194 fixes — FAQ Docker Compose note; simplified OCR deploy prose in multimodal-extraction.md; explicit section anchor IDs in prerequisites-support-matrix.md.

Out of scope (already on main)

• audio-video.md markdown rendering fixes from fix audio-video.md markdown rendering (follow-up to #2179) #2194 are already present on main via the 26.05 merge and follow-up doc PRs.

Follow-up (eng, not this PR)

• LanceDB validation notebook for caption/metadata ingest (per @randerzander review) — cc @jioffe502

Reviewer checklist

• Delete custom-metadata.md; link to metadata filtering notebook from vdbs.md
• Remove chart-caption admonition from support matrix and multimodal-extraction
• Drop 26.05 from image-captioning heading; keep legacy anchor alias
• Rebase/merge main; conflicts resolved

Test plan

• MkDocs build passes
• extraction/custom-metadata.md redirects to vdbs.md#metadata-and-filtering
• FAQ chart-caption answer links resolve to multimodal-extraction.md#charts-and-infographics and #image-captioning
• deployment-options.md offline caption link targets #image-captioning-nim-hardware

[greptile-apps]

Greptile Summary

This PR syncs post-#2179 extraction documentation fixes from the 26.05 branch to main, consolidating the standalone custom-metadata.md page into vdbs.md#metadata-and-filtering and fixing stale anchor references throughout the extraction docs.

• custom-metadata.md removed — content absorbed into vdbs.md#metadata-and-filtering as a compact paragraph plus notebook links; a MkDocs redirect is added, consistent with the fragment-redirect pattern already used by a dozen other redirects in mkdocs.yml; all cross-links in faq.md, integrations-langchain-llamaindex-haystack.md, workflow-agentic-retrieval.md, and the doc-snippet test list are updated.
• Stale anchor chain fixed — #image-captioning-2605 renamed to #image-captioning-nim-hardware on prerequisites-support-matrix.md; a <span id="image-captioning-2605"> alias preserves external bookmarks; deployment-options.md, multimodal-extraction.md, and faq.md retargeted to the new anchor or to verified in-page anchors on multimodal-extraction.md.
• Support-matrix admonition removed — chart-caption scope prose moved to a single sentence under #image-captioning in multimodal-extraction.md; nav renumbered from 7 onward after the single-page "Retrieval & ranking" section is dropped.

Confidence Score: 5/5

Documentation-only reorganisation; no code paths, APIs, or data schemas are changed.

All anchor targets were verified in the post-merge file contents: #charts-and-infographics, #image-captioning, #image-captioning-nim-hardware, and #metadata-and-filtering are all present with the correct explicit IDs. The fragment-redirect pattern for custom-metadata.md is already established by a dozen similar entries in mkdocs.yml. Cross-links in every affected file have been updated consistently, and the deleted file is properly removed from the doc-snippet test list.

No files require special attention.

Important Files Changed

Filename	Overview
docs/docs/extraction/custom-metadata.md	Deleted — content consolidated into vdbs.md#metadata-and-filtering; no content or links are stranded
docs/docs/extraction/deployment-options.md	One-line anchor retarget from #image-captioning-2605 to #image-captioning-nim-hardware, which is the new explicit anchor on prerequisites-support-matrix.md
docs/docs/extraction/faq.md	Two link updates: chart-captioning FAQ now points to verified in-page anchors on multimodal-extraction.md; Docker Compose disclaimer added to env-vars answer
docs/docs/extraction/multimodal-extraction.md	Simplified OCR prose; chart-caption scope sentence added under #image-captioning; cross-links updated to in-page anchors; all targets verified present
docs/docs/extraction/prerequisites-support-matrix.md	Section headings given explicit anchor IDs; #image-captioning-nim-hardware is new canonical anchor; legacy #image-captioning-2605 preserved as a span alias; admonition block removed
docs/docs/extraction/vdbs.md	Metadata-and-filtering section expanded with inline guidance replacing the deleted custom-metadata.md page; old links to that page replaced with notebook links
docs/mkdocs.yml	Nav renumbered after removing the single-page '7. Retrieval & ranking' section; redirect for custom-metadata.md added, consistent with fragment-redirect pattern already used elsewhere
nemo_retriever/tests/test_src_documentation_snippets.py	custom-metadata.md removed from the doc-snippet test list to match the deleted file

Flowchart

%%{init: {'theme': 'neutral'}}%%
flowchart TD
    A["custom-metadata.md\n(deleted)"] -- "redirect" --> B["vdbs.md\n#metadata-and-filtering"]
    C["faq.md"] -- "chart-caption FAQ" --> D["multimodal-extraction.md\n#charts-and-infographics"]
    C -- "image captioning FAQ" --> E["multimodal-extraction.md\n#image-captioning"]
    E -- "NIM & hardware" --> F["prerequisites-support-matrix.md\n#image-captioning-nim-hardware"]
    F -- "legacy alias" --> G["span id=image-captioning-2605"]
    H["deployment-options.md"] -- "offline captioning link" --> F
    I["integrations doc"] -- "filter search link" --> B
    J["workflow-agentic-retrieval.md"] -- "metadata link" --> B

Loading

_{Reviews (11): Last reviewed commit: "docs(extraction): fix caption anchor and..." | Re-trigger Greptile}