docs: remove unpublished root-level pages

[theojin]

What

Removes three root-level pages under docs/ that have no publishing path, plus the single image only they used.

File	Pipeline wiring	Source manifests	samsungtizenos.com
docs/get-started.md	not reached	0 references	404
docs/glossary.md	not reached	0 references	404
docs/menu.yaml	not reached	0 references	—
docs/images/ic_docs_download.png	referenced only by get-started.md	—	—

docs/images/ held that one icon, so the directory goes with it. The identically named icon used by application/dotnet/learn-about-dotnet.md is a separate file under that section's own media/ directory and is untouched.

Why these are unreachable

The only active consumer of master is the tizen.org.v2.docs pipeline, which copies files by slicing the TOCs it reads. I checked all eleven TOC roots and seven fixed-file inputs it uses:

• glossary.md has an entry in docs/application/toc_all.md, but that TOC is only consumed through prefix slices (/application/dotnet/{get-started,guides,reference}/, /application/native/{guides,reference}/, /application/web/{guides,reference,tutorials}/). /glossary.md falls in none of them.
• get-started.md is referenced only from docs/platform/toc_all.md, which no work script reads.
• menu.yaml configured the top navigation of the legacy DocFX site. Nothing in either tizen.org.v2.docs or tizen.org.v2 reads it.

None of the three appears in the committed source manifests (storage/docs/**/{toc.md,toc.json}, storage/documents/**/import.json), and no published document links to any of them.

The legacy site still serves these pages, but it builds from the live branch, which has not moved since 2025-11-20. Content removed from master since then — the IoT, partners, and blog sections, the OpenGL ES guides — is still served there too.

trademarks.md is deliberately kept

It is the third-party trademark attribution page added in 57b3f30 ("Add trademarks page and remove symbols from pages under open-source-project") so individual pages could drop the ®, ™, and © symbols. 144 currently published documents still carry those symbols, and the marks themselves appear throughout the corpus (Wi-Fi 20,883 occurrences, Java 2,235, OpenGL 1,757, Windows 1,702, and so on).

The new site has no equivalent: its terms-of-use covers only Samsung's own intellectual property, and /trademarks returns 404. Its missing publishing path is a separate problem from this cleanup and is filed separately. Its TOC entry in application/toc_all.md and the sentence in application/index.md are both left in place.

Verification

• python3 tools/check_docs.py docs over the whole tree — no findings.
• Ran the tizen.org.v2.docs pipeline twice with --section-only, once against master and once against this branch, over the eight sections that consume application/toc_all.md. Output is identical: same 1,729 files, every Markdown file byte-for-byte equal, and the eight toc.json files differ only in their run timestamps.
• scripts/pages/audit-ssot-coverage.js against the published sitemaps reports work script input missing 0, and none of these files appears anywhere in the report, including wrongly_deleted.
• The published sitemap holds 34,188 URLs and none at /docs/<single-segment> depth, so these pages have no live URL to lose.

Also updated

Inbound references, so the link checker stays clean: the Glossary entry in application/toc_all.md, the get-started entry in platform/toc_all.md, the Glossary sentence in application/index.md, the repository tree in README.md, and the skill's directory-map.md.

[theojin]

Clean removal of three unreachable root-level pages (get-started.md, glossary.md, menu.yaml) and the single icon they used. All inbound references — TOC entries, the index page glossary link, the README tree, and the directory map — are properly cleaned up. trademarks.md is correctly kept since it is still referenced by 144 published documents.

The PR description is thorough: it verifies unreachability through the pipeline TOC slicing, source manifests, and the published sitemap, and confirms the tizen.org.v2.docs pipeline output is byte-identical with and without these files. No issues found.