src/frontend/src/next is 4,123 lines that have never rendered

`src/frontend/src/next/apps/` holds five app views — Desk, Files, Activity, Examples, Settings — totalling 4,123 lines on `main`, committed as WIP in d12f775.

Nothing imports them. Repo-wide grep for `next/apps` returns zero hits outside the folder itself. `vite.config.ts` declares no second input, so the single entry is `index.html` → `/src/main.tsx`, and `main.tsx` does not reach them.

They also have no stylesheet. The views use `dk-`, `cfiles-`, `cact-` and `xex-` class prefixes; `src/styles.css` defines **none** of them. Mounted today they would render as unstyled HTML.

The inverse is true of i18n, and it is worth knowing which way round it is:

| | lines | imports `useT` | has CSS | rendered |
|---|---|---|---|---|
| `src/main.tsx` (shipping panel) | 2,393 | no | yes | yes |
| `src/next/apps/*` | 4,123 | yes, throughout | no | never |

So the only UI a user has ever seen is the one with no translations, and the 451 keys in `src/i18n/en.json` are consumed exclusively by code that does not run. That is the real state of #21, and this issue is why it looks stalled.

Three UIs are now in flight and one renders: the shipping panel, this unmounted desktop rewrite, and the end-user portal that so far exists only as mockups. Deciding what happens to this tree is a prerequisite for the portal, because the portal should not become the third abandoned shell.

Options, in the order I would defend them:

1. **Give it the shell it was written for**, style it, and mount it behind an owner-only route. Recovers the i18n work and closes most of #21.
2. **Harvest and delete.** `Files.tsx` is directly useful to the portal; the rest is admin surface the portal does not want. Keep the logic, drop the tree.
3. **Delete outright** and treat the i18n keys as the spec for translating `main.tsx`.

What should not happen is leaving it where it is. Unmounted code reads as "done" in a file listing and as coverage in a line count, and it is neither.

Start with src/main.tsx and vite.config.ts to confirm the current entry point, then inspect src/next/apps/ and src/styles.css alongside the repo-wide import results described here. Compare the three proposed dispositions and trace the useT imports and en.json consumers. Done means one documented direction is selected and the unmounted tree is no longer left presenting as completed work.

Written by the indexing model from the issue text.