The terminal's crew splits screen is proven by a test: an admin sees runs by model and prompt version, anyone else is not offered it #7936

Intent: every screen the gate ships is covered by a test that would fail if it broke.

Today src/tui/screens/crewsplits.ts (runs by model, by prompt version, model per day, wasted runs per day) is named by no test; test/crew-splits.test.ts covers agentSplits only.

Done:

• [ ] test/tui-crewsplits.test.ts renders the screen for a forge admin over seeded runs on two models and two prompt versions and checks each split's numbers match agentSplits, modelDays and wastedDays.
• [ ] A non-admin's palette does not offer the entry.
• [ ] With no runs it says so in one line instead of an empty table.
• [ ] If a test finds a bug, fix it in src/tui/screens/crewsplits.ts and say so in the pull.

Files: test/tui-crewsplits.test.ts, src/tui/screens/crewsplits.ts (only if a test finds a fault).

Parked by the crew: @scout's run ended without a pull: the pull landed

Done on main already, waiting for you to close it: 1f40c9fa6d321019606c12d24ca28086dd94aeab

Closed: already done on main in 1f40c9f. Closed by the crew's live triage; reopen it to undo, and it stays open.