Cache bug suggestions client-side to avoid redundant refetches

[camd]

Problem

The /jobs details panel Failure Summary tab felt slow. BugSuggestionsModel.get() was a bare fetch with no client-side caching. Because react-tabs unmounts inactive TabPanels (no forceRenderTabPanel), switching away from and back to the Failure Summary tab — or re-selecting a previously viewed job — remounted FailureSummaryTab, which refetched /bug_suggestions/ and re-ran filterGenericFailures on identical data every time. The backend already caches the error summary per job for 24h, so this was pure redundant work on a warm backend.

forceRenderTabPanel was considered and rejected: it eagerly mounts all sibling tabs for every selected job, and SimilarJobsTab fetches on mount, so it would add network load nobody asked for.

Change

Add a bounded, per-session cache in BugSuggestionsModel keyed by jobId, mirroring the existing decisionTaskIdCache pattern in models/push.js:

• Serve cache hits without a network round trip.
• Only cache non-empty results — an empty response means "no failures yet" or "log parsing still in progress", so leaving those uncached lets a later revisit pick up suggestions once the log finishes parsing.
• Cap the cache at 50 entries, evicting the oldest.
• Expire entries after a 1h TTL so a long-lived session eventually picks up a newly-filed bug once the backend's own 24h cache has refreshed.

FailureSummaryTab is untouched — it just benefits from the now cache-aware model.

Why 1h TTL is enough

Measured against the prod replica (30d, 57,486 classifications): 96.5% of classifications reference a bug that was already in use before the job even finished (a lower bound), and of the ≤3.5% "new bug" cases, 97% are filed within 24h of the job during active triage — where the backend's 24h cache is the binding constraint, not the frontend. So staleness from a genuinely new bug is rare, and the 1h TTL comfortably covers it.

Tests

New unit tests in tests/ui/models/bugSuggestions_test.js cover cache hits, the empty-result bypass, TTL expiry, and size-limit eviction.

[codecov-commenter]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 82.60%. Comparing base (740daac) to head (a3b4164).

Additional details and impacted files

@@           Coverage Diff           @@
##           master    #9667   +/-   ##
=======================================
  Coverage   82.59%   82.60%           
=======================================
  Files         622      622           
  Lines       36580    36592   +12     
  Branches     3279     3282    +3     
=======================================
+ Hits        30213    30225   +12     
  Misses       6217     6217           
  Partials      150      150           

☔ View full report in Codecov by Harness.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.
• 📦 JS Bundle Analysis: Save yourself from yourself by tracking and limiting bundle sizes in JS merges.