The operator reads a forge's refused first asks on /forges/refused, linked by one line from /forges #11568

The operator can now read a forge's refused first asks at /forges/refused, and /forges links to it with one line, "Refused first asks".

The earlier tries didn't fail on a test. The last check was stopped for CPU (16.6 cores over a limit of 16), so I changed no code. I re-ran what I could here, and the results were:

• Passing: bun scripts/gate.ts, bun x tsc --noEmit, bun scripts/design-lint.ts and test/forges-refused.test.ts (3 tests) are all clean.
• Not run here: scripts/fast.ts named 1,880 of 2,856 test files for this change. That is too broad to run in this sandbox, so I stopped it and left the full run to the forge's check.
• Push: the push was an empty commit, c9adfade2, because the branch was already up to date and nothing else would re-trigger the check.

Left out: no new behaviour beyond the page and its link.

To try: sign in as admin, open /forges, and follow "Refused first asks" to /forges/refused. The page and its link are covered by test/forges-refused.test.ts, and docs/features/first-ask.md describes them.

If the check stops for CPU again, I'd treat that as the forge's limit rather than a defect in this change.

Closes #11555

The operator can now read a forge's refused first asks at /forges/refused, and /forges links to it with one line, "Refused first asks".

The earlier tries didn't fail on a test. The last check was stopped for CPU (16.6 cores over a limit of 16), so I changed no code. I re-ran what I could here, and the results were:

• Passing: bun scripts/gate.ts, bun x tsc --noEmit, bun scripts/design-lint.ts and test/forges-refused.test.ts (3 tests) are all clean.
• Not run here: scripts/fast.ts named 1,880 of 2,856 test files for this change. That is too broad to run in this sandbox, so I stopped it and left the full run to the forge's check.
• Push: the push was an empty commit, c9adfade2, because the branch was already up to date and nothing else would re-trigger the check.

Left out: no new behaviour beyond the page and its link.

To try: sign in as admin, open /forges, and follow "Refused first asks" to /forges/refused. The page and its link are covered by test/forges-refused.test.ts, and docs/features/first-ask.md describes them.

If the check stops for CPU again, I'd treat that as the forge's limit rather than a defect in this change.

The operator can read refused first asks at /forges/refused, and /forges links to it with one line.

The last check run did not fail on a test. It was stopped because the machine hit 16.7 cores over its limit of 16, with a sh process the busiest. Nothing in this change spawns processes. This is the third such stop, so I pushed an empty commit to re-run the check. If it is stopped for CPU again, the cause is the full suite's shard load on the machine, and a person should look at that.

Locally, before the push:

• scripts/gate.ts is clean.
• tsc --noEmit is clean.
• design-lint is clean.
• test/forges-refused.test.ts passes. I also ran test/forges.test.ts and it passes.
• scripts/fast.ts could not name tests for this diff: it picked 1880 of 2856 files, over the 5-minute budget. I ran only the two files above, and the full suite runs in the forge check.

What I left out is anything beyond reading: the page only lists refused first asks.

To try it, sign in as admin, open /forges, and follow "Refused first asks".

Parked by the crew: the check stayed red through 2 automatic tries