Sentinel: shared:01m23vygyffx5hg2f5vm8r74sr:2jx8fbwpeo is past its budget: trevato/forge: "live triage where there is little to file under > at most 25 issues are filed per tick; the rest are mi… #4534

I gave the setup hook of the bare-repository suite in test/livetriage.test.ts the same 30 s timeout as the 25-issue test. This may not be what was failing on the check machine; the forge's check will show.

• Why I changed it: The setup hook creates a user, which hashes a password. It was on bun's 5 s default, so a loaded check machine could fail it. A hook failure fails every test in the describe, including "at most 25 issues are filed per tick; the rest are missed, then filed".
• Earlier fix: #4485 already gave the 25-issue test itself 30 s. I found no clock or load dependence in src/crew/livetriage.ts, so the verdict depends only on the code under test.
• Why it can't be told from the reports: The sentinel's "failed on N pulls" count may still include pulls that ran before #4485 landed.
• What I ran: bun test test/livetriage.test.ts passes, 34 of 34, in about 1.4 s here. I could not reproduce a failure locally.

I changed no production code and left the rest of the suite's timeouts alone. To confirm, watch whether shared:01m23vygyffx5hg2f5vm8r74sr:2jx8fbwpeo leaves the degraded list.

Closes #4505

I gave the setup hook of the bare-repository suite in test/livetriage.test.ts the same 30 s timeout as the 25-issue test. This may not be what was failing on the check machine; the forge's check will show.

• Why I changed it: The setup hook creates a user, which hashes a password. It was on bun's 5 s default, so a loaded check machine could fail it. A hook failure fails every test in the describe, including "at most 25 issues are filed per tick; the rest are missed, then filed".
• Earlier fix: #4485 already gave the 25-issue test itself 30 s. I found no clock or load dependence in src/crew/livetriage.ts, so the verdict depends only on the code under test.
• Why it can't be told from the reports: The sentinel's "failed on N pulls" count may still include pulls that ran before #4485 landed.
• What I ran: bun test test/livetriage.test.ts passes, 34 of 34, in about 1.4 s here. I could not reproduce a failure locally.

I changed no production code and left the rest of the suite's timeouts alone. To confirm, watch whether shared:01m23vygyffx5hg2f5vm8r74sr:2jx8fbwpeo leaves the degraded list.