fix: keep non-forefront reclaimed requests pending in the single request queue client

[vdusek]

reclaim_request(forefront=False) in the single-mode request queue client discarded the request id from _requests_in_progress and re-cached it, but never re-entered it into _head_requests — the only structure is_empty() and is_finished() read. Once the platform head listing lagged behind the reclaim, both reported the queue finished, AutoscaledPool shut the run down, and the request stayed pending forever while the run reported SUCCEEDED. This is crawlee's ordinary retry path, so any transient request failure near queue exhaustion could silently drop work.

The fix mirrors what add_batch_of_requests already does for non-forefront adds: _head_requests.appendleft(request_id). The forefront=True branch is unchanged. No duplicate risk — _list_head skips ids already in the head, and every crawlee reclaim site operates on a request this client had already popped out of the head.

A regression test, parametrized over forefront, fetches a request through the real path, reclaims it while list_head keeps listing empty, and asserts is_empty()/is_finished() stay False and the request is handed out again. The default case fails without the fix; forefront passed before and after.

Scoped down from the original version of this PR: the shared-client half was made redundant by #1087, and the integration test it added covered the forefront path, which was never broken.

✍️ Drafted by Claude Code

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 92.37%. Comparing base (fcc767a) to head (ba2ae4c).

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #1060      +/-   ##
==========================================
+ Coverage   92.31%   92.37%   +0.06%     
==========================================
  Files          51       51              
  Lines        3317     3318       +1     
==========================================
+ Hits         3062     3065       +3     
+ Misses        255      253       -2     

Flag	Coverage Δ
e2e	34.71% <0.00%> (-0.02%)	⬇️
integration	56.57% <100.00%> (+0.01%)	⬆️
unit	88.75% <100.00%> (+1.81%)	⬆️

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.