Bug 1499551 - Bound jobs-view memory: per-repo push cap + idle-timeout polling (part 2/3)

[camd]

Part 2 of 3 addressing Bug 1499551 (jobs-view memory growth over time). Expanded per team discussion to also bound background-tab lifetime.

1. Per-repo rolling push cap

On a watched tree the 60s poll appends pushes and their jobs with no eviction, so pushList/jobMap/decisionTaskMap and the per-<Push> job state grow unbounded. Now:

• Rolling cap on retained pushes: 200 for autoland, 100 for other repos (getRepoPushCap).
• Cap-polling-only: the cap is a floor; explicit "get more pushes" loads still raise the limit, so users can browse deeper history. Polling-driven growth is evicted back to the cap (oldest first), pruning those pushes' jobs — which is what reclaims the memory.
• The selected push and any push with a pinned job are never evicted.

2. Stop polling forgotten background tabs (idle timeout + keep-alive)

• A tab left in the background (Page Visibility hidden) stops the push/job poll after 2 hours of continuous inactivity. Bringing it back to the foreground resumes polling (with one catch-up poll) and resets the clock.
• A keep-alive toggle (a coffee mug) in the jobs-view toolbar extends that window to 12 hours for a tab you intend to leave open. Per-tab, default off, not persisted.
• Only the push/job poll stops; app-level intervals (deployed-version check, notifications) keep running.

New pieces: pollControlStore (per-tab keepAlive/pollingPaused), usePollingLifecycle hook (interval + visibility idle logic, extracted from PushList for isolated testing), and KeepAliveButton.

Testing

• Per-repo cap: getRepoPushCap, enforcePushLimit eviction + selected/pinned protection + jobMap pruning, floor applied on fetchPushes, trim after pollPushes.
• Idle timeout: usePollingLifecycle with fake timers + simulated visibility changes — stops after the limit while hidden, resumes + catch-up on refocus, refocus resets the clock, keep-alive extends to 12h.
• pollControlStore and KeepAliveButton (toggle + paused indicator).
• Full frontend unit suite passes; lint clean.

Part 3 (#9706) is stacked on this branch.

[codecov-commenter]

Codecov Report

❌ Patch coverage is 93.93939% with 2 lines in your changes missing coverage. Please review.
✅ Project coverage is 82.67%. Comparing base (6a820e6) to head (8190206).
⚠️ Report is 2 commits behind head on master.

Files with missing lines	Patch %	Lines
ui/shared/stores/pushesStore.js	93.93%	1 Missing and 1 partial ⚠️

Additional details and impacted files

@@           Coverage Diff           @@
##           master    #9705   +/-   ##
=======================================
  Coverage   82.66%   82.67%           
=======================================
  Files         626      626           
  Lines       36753    36784   +31     
  Branches     3348     3338   -10     
=======================================
+ Hits        30383    30412   +29     
- Misses       5988     5989    +1     
- Partials      382      383    +1     

☔ 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.