fix: avoid crawling handled requests again after migration in large request queues

[vdusek]

In single-access mode, a request that the batch add reports as already handled is recorded as handled locally and kept out of the cache and the head, so fetch_next_request doesn't hand it out again. Before, this hit runs with more than 10,000 requests after a migration, reboot or resurrection: re-added start URLs and re-discovered links were crawled twice.

The shared client also takes the platform's handled flag into its dedup cache, for consistency.

Closes: #1185

✍️ Drafted by Claude Code

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 93.07%. Comparing base (290af90) to head (7585abd).

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #1189      +/-   ##
==========================================
+ Coverage   93.00%   93.07%   +0.06%     
==========================================
  Files          54       54              
  Lines        3672     3680       +8     
==========================================
+ Hits         3415     3425      +10     
+ Misses        257      255       -2     

Flag	Coverage Δ
e2e	31.65% <0.00%> (-0.07%)	⬇️
integration	51.73% <77.77%> (+0.05%)	⬆️
unit	89.83% <100.00%> (+0.10%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

☔ 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.