Add a CI job validating against GHC's JavaScript backend

[zlonast]

fix: #12293

Known limitations (upstream, not Cabal bugs)

Note

These are the reasons the draft waits for GHC 9.12.5 / 9.14.2 / 10.0.
Each item is a GHC-side gap that is already known/fixed upstream; none of
them are regressions introduced here.

1. cabal-testsuite cannot run on the JS target — it executes every test
   script inside a GHCi session (Test.Cabal.Server), and the 9.12.2 cross
   bindist is "not built for interactive use". This includes the new
   PackageTests/JS/* tests (they are skipUnlessJavaScript, so native CI
   runs keep skipping them).
2. cabal-install unit tests fail to load — hackage-security uses
   StaticPointers, and the JS backend mis-links static pointer tables
   (ReferenceError: h$$... is not defined at load time). The JS-built
   cabal executable itself links and runs fine.
3. no-thunks-test needs unpackClosure# — introspection primop not
   provided by the JS backend.
4. rpmvercmp suite needs a C FFI routine — its reference implementation
   is C; bridging it through the emscripten heap would defeat the purpose of
   the cross-check.
5. RTS gaps worked around here (h$fchmodat, h$CLOCK_*,
   h$getNumberOfProcessors, cryptohash-sha256 FFI) — tracked in the
   polyfills file, removable when the runtime catches up.

---

Template Α: This PR modifies behaviour or interface

Include the following checklist in your PR:

• Patches conform to the coding conventions.
• Any changes that could be relevant to users have been recorded in the changelog.
  • Is the change significant? If so, remember to add significance: significant in the changelog file.
• The documentation has been updated, if necessary.
• Manual QA notes have been included.
• Tests have been added. (Ask for help if you don’t know how to write them! Ask for an exemption if tests are too complex for too little coverage!)