Add device code review skill and refresh CI

[Binyang Li (Binyang2014)]

Summary

• add a reusable device-code review skill for CUDA/HIP changes
• update the development container base image to CUDA 13.0
• run lint jobs on Ubuntu 24.04
• remove the CUDA compatibility directory from the mscclpp-lang workflow's LD_LIBRARY_PATH
• restore the original Azure build path on remote coverage VMs so lcov can read source files
• tolerate lcov 2.x inconsistent coverage metadata and use accurate summary output
• disable the ROCm 7.2 code coverage matrix entry
• skip host-side atomic signaling tests in SM90 and ROCm Debug coverage runs

Testing

• H100 Debug coverage with the filters: unit tests 4.6 seconds, 2-rank tests 53.2 seconds, 4-rank tests 55.4 seconds
• verified the filter condition remains disabled for CUDA SM80 and enabled for CUDA SM90 and ROCm
• A100 coverage continues to exercise the excluded paths
• ran ./tools/lint.sh

[azure-pipelines]

Azure Pipelines:
There may be pipelines that require an authorized user to comment /azp run to run.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 67.53%. Comparing base (be9126c) to head (3ca5207).
⚠️ Report is 53 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #862      +/-   ##
==========================================
+ Coverage   59.62%   67.53%   +7.91%     
==========================================
  Files          57       61       +4     
  Lines        5278     5844     +566     
==========================================
+ Hits         3147     3947     +800     
+ Misses       2131     1897     -234     

Flag	Coverage Δ
cuda-80	62.56% <ø> (+2.32%)	⬆️
cuda-90	67.73% <ø> (?)
rocm-gfx942	60.00% <ø> (-3.43%)	⬇️

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.