test(pt_expt): apply fast AOTI config to generated pt2 fixtures

[njzjz-bot]

Problem

• PR test(pt_expt): speed up AOTInductor UT compilation and fix hardcoded device #5394 applies fast AOTInductor settings from source/tests/pt_expt/conftest.py, but the C/C++ test setup generates .pt2 fixtures by running source/tests/infer/gen_*.py directly.
• Those generation scripts do not load pytest conftest.py, so the slow C++ fixture compilation path does not benefit from the settings.

Change

• Move the fast test-only AOTInductor settings into a shared tests.pt_expt.aoti_config helper.
• Apply the helper from pt_expt pytest conftest.py.
• Apply the same helper from source/tests/infer/gen_common.py, which is used by all C/C++/LAMMPS .pt2 fixture generation scripts.

Notes

• This is a draft PR to measure CI timing impact.
• Local verification is limited to python3 -m py_compile and git diff --check; this environment does not have PyTorch installed.

Authored by OpenClaw (model: custom-chat-jinzhezeng-group/gpt-5.5)

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 82.20%. Comparing base (69409c5) to head (42b6b46).
⚠️ Report is 3 commits behind head on master.

Additional details and impacted files

@@            Coverage Diff             @@
##           master    #5546      +/-   ##
==========================================
- Coverage   82.22%   82.20%   -0.02%     
==========================================
  Files         892      892              
  Lines      101548   101562      +14     
  Branches     4242     4243       +1     
==========================================
+ Hits        83493    83494       +1     
- Misses      16754    16765      +11     
- Partials     1301     1303       +2     

☔ 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.