chore: Disable the TraceMalloc allocator

[coastalwhite]

No description provided.

[codecov]

Codecov Report

All modified and coverable lines are covered by tests ✅

Project coverage is 79.76%. Comparing base (03df8e4) to head (5e2e711).
Report is 35 commits behind head on main.

Additional details and impacted files

@@            Coverage Diff             @@
##             main   #21231      +/-   ##
==========================================
- Coverage   79.76%   79.76%   -0.01%     
==========================================
  Files        1594     1593       -1     
  Lines      228142   228119      -23     
  Branches     2600     2600              
==========================================
- Hits       181974   181949      -25     
- Misses      45571    45573       +2     
  Partials      597      597              

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

[coastalwhite]

@itamarst, pinging you since you added this and were involved with python/cpython#128679.

I have basically removed the tracemalloc infrastructure here while leaving the MemoryUsage fixture on the Python side for a possible future implementation. It is really useful to have a good (peak) memory usage tracking solution for the tests and CI, but this is not really it. We would be open to a different implementation relying on psutils or something alike. Although, I am not sure how compatible that is with pytest.

[itamarst]

Yep, totally understand. Another possibility is memray, which works with a completely different mechanism, so I will look into that as a replacement.

Is there an upstream bug filed with CPython about this?

[coastalwhite]

Yep, totally understand. Another possibility is memray, which works with a completely different mechanism, so I will look into that as a replacement.

Is there an upstream bug filed with CPython about this?

We did not reproduce outside the CI so I was unable to make an MRE. I also did see memray, not 100% about their approach, though.