test: concurrent page allocator

[jpnurmi]

In #974, the originally reported crash was caused by the old Unix page allocator spinlock, whose unlock lacked release semantics. On ARM64 this could corrupt allocator state and produce overlapping allocations; sentry_value_set_by_key_n was where the resulting object corruption surfaced.

The synchronization was corrected in #1446. This PR adds the missing regression test: multiple threads allocate concurrently, write unique markers, and verify that every allocation remains independent.

Temporarily restoring the non-atomic pre-#1446 unlock reproduces the race under TSAN, while the corrected implementation passes:

WARNING: ThreadSanitizer: data race
  Write of size 4 by thread T7:
    #0 sentry__page_allocator_alloc src/sentry_unix_pageallocator.c:130

  Previous atomic read of size 4 by thread T3:
    #0 sentry__page_allocator_alloc src/sentry_unix_pageallocator.c:98

  Location is global 'g_lock' of size 4

SUMMARY: ThreadSanitizer: data race in sentry__page_allocator_alloc

https://github.com/getsentry/sentry-native/actions/runs/33185811927/job/98898403787?pr=2035

Ref: #974

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 74.36%. Comparing base (719803e) to head (62bc7ce).

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #2035      +/-   ##
==========================================
+ Coverage   74.21%   74.36%   +0.15%     
==========================================
  Files         104      104              
  Lines       26627    26627              
  Branches     4844     4844              
==========================================
+ Hits        19762    19802      +40     
+ Misses       5519     5477      -42     
- Partials     1346     1348       +2     

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.