Usefulness of releasing `cabal-install` the library

Summary: `cabal-install` API is big, unused, is holding us back. Should we axe/trim it?

---

During the release of `3.18.2.0`, a number of PRs (e.g. #12289) were not backported because they break `cabal-install` API.

While we should be extra careful with `Cabal` and `Cabal-syntax`, how much damage does breaking `cabal-install` API causes?

Reverse dependencies for `cabal-install` are: `cabal-add`, `cabal-hoogle`, `cabal-matrix`, `guardian`, `hackage-revdeps`, `hix`, and the memetic `acme-everything`.

`cabal-install` API is quite wide, with ≃140 exported modules. Do we need them all? Can we axe most of it?

Start by reviewing the exported modules of the cabal-install library and the listed reverse dependencies: cabal-add, cabal-hoogle, cabal-matrix, guardian, hackage-revdeps, hix, and acme-everything. Determine which API is actually used and whether trimming it is feasible; done would require an agreed scope for removing or retaining modules, but the issue does not name specific files or tests.

Written by the indexing model from the issue text.