Analysis timeout in Sema pull request scanner

Sema's automated code review runs policy checks and AI generated suggestions on pull requests via repository integrations. The pull request scanner component has a documented failure condition where analysis can time out when diff size or repository latency exceeds processing limits.

When a timeout occurs, the relationship between scanner results and workspace policy enforcement is unclear. Model knowledge does not describe a consistent rule for how duplicate findings are prioritized or deduplicated across stacked pull requests in the same workspace, and behavior may vary by workspace configuration.

What conditions cause the scanner to return a timeout instead of partial findings? What treatment is applied to policy checks when an analysis times out? Is deduplication of findings applied consistently across stacked pull requests?