Benchmarks failing in CI

The benchmarks in CI fail every time now.

Example: https://github.com/JuliaPy/PythonCall.jl/actions/runs/19360980161/job/55392621221?pr=716

The error is something about a commit not existing.

@MilesCranmer any idea what the fix is?

Start with the linked GitHub Actions run and inspect the benchmark CI configuration and its commit reference. Reproduce the failure, determine why the referenced commit is unavailable, and verify that the benchmarks complete successfully in CI.

由索引模型根据 Issue 内容生成。