Consensus lookups in the extension have no timeout and run one finding at a time #111

Found while merging #89 into #109.

After the model call, which #109 caps at 25 s, the service worker runs verifyFindingsWithConsensus. For each finding it awaits its own fetch('https://api.consensus.app/v1/search'), one after another, with no AbortSignal and no overall limit. A slow Consensus API therefore holds the audit open indefinitely. MV3 also kills a worker whose fetch takes more than 30 s to respond, which is the reason #109 added the 25 s model timeout.

Possible fix: AbortSignal.timeout per request, and Promise.all across findings (or a small concurrency cap). Keep the current behavior of treating a failed lookup as "no badge". The CLI already sets timeout=10 in fetch_from_consensus_api.

🤖 Generated with Claude Code