fix: stop the Windows CI job timing out on integration tests - #34

No actionable issues found.

Overall this change is a focused and well-scoped fix for a concrete CI problem. The approach of conditionally extending the job-level timeout based on the runner OS is reasonable and avoids globally inflating timeouts for the Linux and macOS jobs, which already complete comfortably within the existing limit. The timeout value chosen for Windows appears proportionate to the integration test workload, and the if condition is written clearly so future maintainers can understand and adjust it without confusion.

No issues were found in the changed code. The workflow syntax is valid, the conditional expression is well-formed, and the modification does not introduce any risk of bypassing other intended guardrails (such as cancellation of hung jobs or visibility into long-running steps). It also leaves the existing per-step timeouts untouched, which preserves the layering between job-level and step-level timeout policies.

That said, I would recommend a couple of follow-ups outside this PR to make the fix more durable. First, it would be worth commenting why Windows specifically needs the extra time — whether it is slower runner performance, antivirus overhead, or particular integration tests that are inherently slower — so the next person to touch this line does not accidentally revert it. Second, since integration tests seem to be the driver here, consider whether those tests can be parallelized, sharded, or split into a separate workflow so the CI critical path stays fast for everyone. The change as written is sound and safe to merge.