`bun scripts/home-eval.ts` scores the owner's home against ground truth: which of the five facts that mattered the page states, and which it states wrong #5805

Part of #5735's validation loop. The cold-read agent is the steward's to run; this gives the loop a deterministic half that the crew can run on every feed change.

Done when:

• [ ] The script reads the page as text with #5765's reader. It builds ground truth from the forge's own API: landed pulls and releases in the window, items waiting on the owner, spend against the forge cap, builders idle and why, and repositories with no commit in 3 days.
• [ ] It prints each of those facts as stated, missing or wrong, with the page's words beside it. Then it prints the coverage (n of 5), the errors, the characters and the fetch time.
• [ ] docs/HOME-EVAL.md gives the method in the owner's words from #5735 and keeps a table of results. Each feed pull adds a row.
• [ ] A test runs it against a seeded forge where one fact is wrong and checks that it says so.