fix(tests): scale daemon runtime sanitizer timeout

[bmcnaboe]

What does this PR do?

Uses sanitizer-aware timing budgets for the blocking-Git daemon cancellation regression while preserving the existing native timing budgets and cancellation/cleanup behavior.

Root cause

The ASan/UBSan runner is about 421 MB and can take longer than five seconds to reach main after being copied. The test therefore failed at its readiness backstop before exercising cancellation.

Checklist

• Every commit is signed off (git commit -s)
• Tests pass locally (make -f Makefile.cbm test)
• Lint passes (make -f Makefile.cbm lint-ci)
• New behavior is covered by the existing regression

Validation

• scripts/test.sh --suites daemon_runtime: 45 passed
• real pre-commit hook: 7,555 passed, 8 skipped
• scripts/test.sh: 7,555 passed, 0 failed, 8 skipped
• production watchdog and security guards passed

Fixes #1738

[github-actions]

Thanks for opening this — it has been seen, and it is queued.

This note is automated, but it is not a brush-off: it exists so you know where your PR stands instead of having to guess from silence.

Current review status: working through a backlog. 0.9.1-rc.1 is out, so the release freeze that held reviews is over — but it left a large queue of open pull requests behind it, and we are reading through them oldest-first. The background is in discussion #1144.

What that means for this PR, concretely:

• It will not be closed for inactivity. No stale bot touches pull requests here.
• It may still sit a while before a human reads it. That is on us, not on you.
• Older PRs are read first, so a recent one is not being skipped — it is behind a queue.

Things that will genuinely speed it up whenever review does happen:

• Keep it rebased on main — the tree is moving quickly right now, and a conflicting branch cannot be reviewed as the diff you intended.
• Get CI green, or say which failures you believe are pre-existing.
• Keep the change to one claim. Bundled features and refactors get split before they get merged, which costs you a round trip.
• Every commit needs a sign-off (git commit -s) — CI enforces DCO.

If this fixes a bug, a reproduction we can run is worth more than a description of the symptom.

Thanks for contributing, and sorry in advance for the wait.

[DeusData]

Thank you for isolating the sanitizer-specific startup budget while preserving the native timing and cancellation contracts. The contribution queue is quite full, so review may take a little time. We have the PR routed and will come back with grounded feedback as soon as possible.

[DeusData]

Approved. Timing-budget changes to tests get read carefully here, because the usual version of this change is tuning a racy assertion until it passes. Yours is not that, and the diff shows why in one glance.

You scaled the precondition and left the measurement alone. CHILD_CANCEL_BOUND_MS (3000) and CHILD_CLEANUP_BOUND_MS (5000) are untouched — those bound the behaviour actually under test, and they stay exactly as strict as they were. Only CHILD_READY_BOUND_MS and REQUEST_TIMEOUT_MS move, and only under #if CBM_SANITIZED. That split is the whole difference between a legitimate budget and a weakened test: if the cancellation bound had moved too, this would be a different conversation.

And the comment says which is which, right where the next person will meet it: "This readiness wait is a liveness backstop, not the behavior under test." That sentence is what stops someone in six months treating all four constants as interchangeable knobs.

The root cause is a slow deterministic thing, not a racy one. A 421 MB instrumented runner being copied and then taking more than five seconds to reach main is latency, not a race — the process gets there, it just takes longer under instrumentation and more still while the parallel gate is loaded. Waiting longer for something that will definitely happen is fine; waiting longer for something that might not is what we do not do.

Using the compile-time CBM_SANITIZED discriminator rather than sniffing at runtime is right too — the budget follows the build that actually needs it.

One thing worth naming rather than changing: a 60-second readiness backstop means a genuine hang now costs 60 seconds to surface instead of 5. That is the correct trade for a liveness backstop, but it is a real cost if this ever starts failing for a different reason, and worth remembering then.

Your CI is the same story as #1741

All five reds are one cascade, and none of them is yours. security / codeql-gate ran 16:26:30 → 17:12:17 — 45 minutes 47 seconds — which is the gate's old wait budget expiring rather than a finding. lint / lint and lint / lint-mem both record The operation was canceled with zero warnings of their own, and analyze and ci-ok fall out of the same run. That is the identical signature to your other PR.

The budget was raised from 45 to 150 minutes in #1960, merged yesterday. I will update this branch and re-run before merging — main also moved three times yesterday. Nothing needed from you.

Two small, well-argued infrastructure fixes sitting two weeks behind a gate timeout that had nothing to do with either of them. Sorry about that.