Reduce duplication & accurcy for RAG index collection pages

[lisa-tarbo]

Background to changes

While working on adding search/retrieval to RAG developer documentation dimagi/open-chat-studio#1716, I noticed inaccuracies and duplication in the user guides on this topic

Improvements made

1. Versioning.md contained information about collection files for published chatbots. This is now merged into Collections concept page. Reduced duplication
2. Gemini was missing from the LLM providers that provide local indexed collections
3. Clarify doc source only supports 2 types (Github and Confluence) early in page as it impacts the users decisions

Also formatting, grammar and readablity

[claude]

Docs review

Nice consolidation — the versioning.md → collections/index.md merge cleanly removes the duplication, the drift-detection note and existing-published-bots warning were preserved verbatim, and the Gemini addition is consistent with the rest of the docs. I verified all changed cross-references resolve (indexed.md#document-sources-for-indexed-collections, index.md#collections-and-published-chatbots, document_sources.md#monitoring-sync-status) and no lingering references to the removed anchors (#document-sources, #what-is-frozen-and-what-is-live) remain anywhere in docs/.

A few items, left as inline comments:

• ⚠️ collections/index.md (lines 5–8) — the new - markers on the definition-list terms will likely break def_list rendering (every other def-list on the site uses a bare term line). Please preview locally before merging — this is the one blocking item.
• collections/indexed.md (line 65) — double space in the providers list (nit).
• how-to/document_sources.md (line 6) — "external platform" collides with the channel = "platform" terminology used elsewhere; suggest "external system" / "external document source" for consistency.

Minor / non-blocking: indexed.md frontmatter title (singular) no longer matches the H1 (plural, now "...(for RAG applications)") — good chance to align while the line's already being touched.

[claude]

Documentation review

The consolidation of the versioning/collections content is well done — no information is lost, both admonitions and the 2026-06-03 date carry over intact, the heading rename is reflected everywhere, and all the updated anchors (#document-sources-for-indexed-collections, #collections-and-published-chatbots, etc.) resolve. Adding Gemini and clarifying the two document-source types up front are good, user-facing improvements.

One thing to fix before merge, plus a couple of optional nits.

🔴 Fix before merge

• docs/concepts/collections/index.md (lines 5–6): the two collection entries lost their definition-list markup and now render as a single run-together paragraph (no blank line between the lines). Left an inline suggestion to convert them to a bulleted list.

🟡 Suggestions (optional)

• Provider ordering is inconsistent between the two companion pages: indexed.md says "Confluence … or GitHub" while document_sources.md lists "GitHub and Confluence". Picking one canonical order would read as more deliberate.
• index.md line 16: the colon-into-bullets works, but the media-collection bullet's trigger (adding/removing files) isn't really "via a document-source sync" — only the second bullet is. Minor clarity nit.

👍 Nice

• Promoting the two troubleshooting items in document_sources.md to real ### headings makes them TOC- and anchor-addressable.

No code examples to verify (docs-only). Nothing broke in the cross-references.