Fix ring all-reduce in-place corruption at gpu_num ≥ 3

[chhwang]

Fix ring all-reduce in-place corruption at gpu_num ≥ 3

In-place ring all-reduce mutates input before the next iteration's
send when output aliases input. At TP=2 the single iteration hides
the bug; at gpu_num ≥ 3 every subsequent send carries stale data.

Fix: detect buffer aliasing (output.ref()->buffer()->id() == input.ref()->buffer()->id()) and copy input into a private tensor
before the ring loop.

Also sorted includes in ops_all_reduce_test.cpp to satisfy
clang-format.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 86.78%. Comparing base (75de9b4) to head (8d4625d).
⚠️ Report is 1 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #253      +/-   ##
==========================================
- Coverage   86.88%   86.78%   -0.10%     
==========================================
  Files         125      127       +2     
  Lines        6113     5993     -120     
==========================================
- Hits         5311     5201     -110     
+ Misses        802      792      -10     

☔ 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.