Investigate pref regression in TestFindManyAndEmptyCursor benchmark. FindManyAndEmptyCursor dropped ~35% back in late July:
https://spruce.mongodb.com/task/mongo_python_driver_perf_tests_perf_6.0_standalone_63d957c2137cec66821d4d1669ff7a24f4c4f6f0_24_09_11_15_22_22/trend-charts?execution=0&sortBy=STATUS&sortDir=ASC

This regression is present in both the ssl and non-ssl benchmark so it seems real. Unrelated to this ticket itself but we should also investigate why we never got an alert for this.

So far I can't reproduce the regression locally on macOS. I scheduled a few benchmark runs on the commits in between https://github.com/mongodb/mongo-python-driver/commit/0f5349e2e02716611a2fe4beb0605f74fcf63fe3 (PYTHON-4572) and https://github.com/mongodb/mongo-python-driver/commit/105393109e891b35f64eec3df11955f2cdc2e81d (PYTHON-4533) to narrow down the search.