fix(docs): cherry-pick v2.6.0 Sphinx and version switcher fixes

[shubhadeepd]

Summary

Cherry-picks the docs fix from PR #664 onto main.

This updates the Sphinx docs build configuration, fixes documentation cross-reference and style issues, and updates the version switcher manifest for the v2.6.0 documentation line.

Notes

• Source commit: f66cdec10d5e5c2d4a77d4dde4a4b4dea37797fa
• Cherry-pick commit: 875ab0a039da8869567fd82c5fadea5bfd268905
• Resolved one conflict in docs/versions1.json by keeping main's 2.5.1 entry and adding the new name fields expected by the version switcher.

Validation

• uv run --group docs sphinx-build docs docs/_build/html

Summary by CodeRabbit

• Documentation

  • Enhanced Agentic RAG documentation with improved formatting and structured pipeline overview.
  • Updated performance benchmarking guide with refined CLI guidance and clearer configuration instructions.
  • Improved VLM feature documentation with clarified explanations and updated external references.
  • Updated release notes with corrected links and refined content descriptions.
  • Expanded Developer Guide navigation with new deployment configuration entry.

• Chores

  • Updated version manifest configuration to designate latest release as preferred.

[copy-pr-bot]

Auto-sync is disabled for draft pull requests in this repository. Workflows must be run manually.

Contributors can view more details about this message here.

[coderabbitai]

📝 Walkthrough

Walkthrough

This PR updates the NVIDIA RAG Blueprint documentation with a build-system enhancement and comprehensive content improvements. The Sphinx configuration now manages the version manifest at build time via hooks, the version list marks 2.6.0 as preferred, and feature documentation for agentic RAG, VLM, performance benchmarking, and release notes receives clarification, restructuring, and link updates to external resources.

Changes

Documentation Build System and Content Updates

Layer / File(s)	Summary
Sphinx build hook for version switcher management docs/conf.py	Introduces _copy_versions_json() helper and setup(app) Sphinx hook to copy versions1.json at build time with environment-variable support for switcher URL, removes static versions1.json from html_extra_path.
Version metadata and navigation structure docs/versions1.json, docs/index.md	Adds "preferred": true to 2.6.0 version entry; adds Nemotron 3 Super Deployment navigation entry and adjusts Multimodal Retriever link label in Developer Guide.
Agentic RAG documentation restructuring docs/agentic-rag.md	Reformats "It includes" and "Limitations" with consistent Markdown emphasis; restructures "Architecture Overview" into numbered stages; adds "Enable per request" anchor and image block with width in streaming UI subsection.
VLM documentation clarifications and deployment guidance docs/vlm.md	Updates "Integrated multimodal experience" terminology, restructures pipeline and reasoning-mode explanations, clarifies environment-variable parameter setup, expands Docker Compose guidance with service confirmation command, refines multimodal retriever references.
Release notes link and wording updates docs/release-notes.md	Updates agent skills and RAG performance tooling links to GitHub URLs across releases 2.6.0–2.4.0; changes "Easy model switches" to "Direct model switches" in Release 2.4.0; rewords Kubernetes values.yaml known issue.
Performance benchmarking documentation consolidation docs/performance-benchmarking.md	Updates rag-perf introduction with GitHub URL, adds preset copy-locally tip, refines workflow descriptions, expands stdout output guidance, substantially rewrites "Query inputs" section to clarify mutual exclusivity and formats, reformats configuration table, updates related topics to absolute GitHub URLs, and restructures output layout with explicit aiperf.enabled variants.

Estimated code review effort

🎯 2 (Simple) | ⏱️ ~12 minutes

Suggested reviewers

• nv-pranjald

Poem

🐰 Docs grow clearer, versions now preferred,
Hook copies manifest at build, not deferred;
Agentic, VLM, benchmarks refined—
Each page restructured, content aligned! ✨

🚥 Pre-merge checks | ✅ 4 | ❌ 1

❌ Failed checks (1 warning)

Check name	Status	Explanation	Resolution
Docstring Coverage	⚠️ Warning	Docstring coverage is 25.00% which is insufficient. The required threshold is 80.00%.	Write docstrings for the functions missing them to satisfy the coverage threshold.

✅ Passed checks (4 passed)

Check name	Status	Explanation
Description Check	✅ Passed	Check skipped - CodeRabbit’s high-level summary is enabled.
Title check	✅ Passed	The PR title clearly and accurately describes the main change: cherry-picking Sphinx and version switcher documentation fixes from v2.6.0 into main.
Linked Issues check	✅ Passed	Check skipped because no linked issues were found for this pull request.
Out of Scope Changes check	✅ Passed	Check skipped because no linked issues were found for this pull request.

_{✏️ Tip: You can configure your own custom pre-merge checks in the settings.}

✨ Finishing Touches

📝 Generate docstrings

• Create stacked PR
• Commit on current branch

🧪 Generate unit tests (beta)

• Create PR with unit tests
• Commit unit tests in branch agent/cherry-pick-pr-664-to-main

---

_{Comment @coderabbitai help to get the list of available commands and usage tips.}

[coderabbitai]

Actionable comments posted: 4

🤖 Prompt for all review comments with AI agents

Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.

Inline comments:
In `@docs/agentic-rag.md`:
- Around line 66-67: Replace fragile plain-fragment links that target the
existing MyST label "(enable-per-request)=" with Sphinx-style cross-references:
find occurrences of "(`#enable-per-request`)" (and any other "`#enable-per-request`"
fragments) and change them to use the MyST/Sphinx ref role syntax
{ref}`enable-per-request` so references resolve reliably across builders; update
all matching links in this document (e.g., near the earlier link around line
~32) to the new {ref}`enable-per-request` form.

In `@docs/conf.py`:
- Around line 57-62: Add missing type hints to the Sphinx extension functions:
annotate setup as def setup(app: Any) -> None and the callbacks as def
_on_builder_inited(_app: Any) -> None and def _on_build_finished(_app: Any, exc:
Optional[Exception]) -> None, and import Any and Optional from typing (or use
more specific Sphinx types if available, e.g., Sphinx from sphinx.application)
so every function signature has explicit type annotations.

In `@docs/performance-benchmarking.md`:
- Around line 163-170: The fenced JSONL example block is missing blank lines
before and after the triple-backtick fence which triggers MD031; fix the
`.jsonl` list item by ensuring there is a blank line immediately before the
opening ```json fence and a blank line immediately after the closing ``` fence
so the code block is separated from surrounding list text (adjust the fenced
JSONL example shown under the `.jsonl` bullet accordingly).
- Around line 323-332: The three fenced examples showing output layouts (the
blocks containing "run_<ts>/{report.md, results.csv, results.json, profiling/,
aiperf_rag_on/}", "run_<ts>/{profile_report.md, profile_results.json,
profiling/}", and the nested tree starting with "run_<ts>/") need a blank line
before and after each fence and must declare a language; change each fence from
``` to ```text and ensure there is an empty line separating the surrounding
paragraph and the code fence so the blocks become properly fenced (e.g., add a
blank line, replace ``` with ```text, paste the snippet, then add a blank line
after the closing fence).

🪄 Autofix (Beta)

Fix all unresolved CodeRabbit comments on this PR:

• Push a commit to this branch (recommended)
• Create a new PR with the fixes

---

ℹ️ Review info

⚙️ Run configuration

Configuration used: Path: .coderabbit.yaml

Review profile: ASSERTIVE

Plan: Enterprise

Run ID: 5d340e15-94f1-422e-b01d-9c4e4513a88e

📥 Commits

Reviewing files that changed from the base of the PR and between 0f216e1 and 875ab0a.

📒 Files selected for processing (7)

• docs/agentic-rag.md
• docs/conf.py
• docs/index.md
• docs/performance-benchmarking.md
• docs/release-notes.md
• docs/versions1.json
• docs/vlm.md