Hugo and the Hextra fork, instead of MkDocs Material and mike

[tannevaled]

One of the fleet's 85 MkDocs sites, moved to Hugo and the
Hextra fork pinned at v0.16.0.

Converted by mkdocs2hextra,
not by hand. Its own report:

note: built from a MODIFIED tree at b7753702cffe -- this reading is not reproducible from any commit.
go-webengine docs: 4 pages in the nav
  wrote .github/workflows/docs.yml                     5706 bytes
  wrote README.md                                      1237 bytes
  wrote content/en/_index.md                           6021 bytes
  wrote content/en/architecture.md                     4998 bytes
  wrote content/en/fidelity.md                         5334 bytes
  wrote content/en/roadmap.md                          4043 bytes
  wrote hugo.yaml                                      2625 bytes
carried over: 0 callouts, 0 details, 0 tab sets, 7 links, 0 attribute lists dropped, 0 snippets left, 0 mermaid

still there, and the old site builds from them — remove in the same commit:
  mkdocs.yml
  requirements.txt
  docs/ (5 entries — the pages are now under content/en/)

this site was versioned with mike: the Hugo build keeps a path per version (params.docs.version = "main")

the nav's order	becomes Hugo's weight — nothing else in either system knows it
every link to a .md file	becomes {{< relref >}}: Hugo serves /page/, not /page.md, so every one was broken by the move, and a relref fails the build when its target is missing
each documented version	keeps a path of its own (/docs/<version>/, latest following the newest), which is what mike was doing
MkDocs Material + mike	needed a Python toolchain, a requirements.txt and a pip install in CI to render a static site for a pure-Go organisation

⛔ It was built locally before this branch was pushed, with
--panicOnWarning: nothing is committed for a site that does not build.

⛔ The branch is cut from origin/main, not from the local one.

🤖 Generated with Claude Code