CI(benchmark): run benchmark on codspeed-macro for wall time (#858)

The previous results are not reliable, due to
CodSpeedHQ/pytest-codspeed#37. The new method
is good.

See
https://docs.codspeed.io/instruments/walltime/#usage-with-github-actions

<!-- This is an auto-generated comment: release notes by coderabbit.ai
-->

## Summary by CodeRabbit

* **Chores**
* Updated the benchmark workflow to run only for the "deepmodeling"
repository owner and switched the runner environment to a custom setup.

<!-- end of auto-generated comment: release notes by coderabbit.ai -->

Author: njzjz

.github/workflows/benchmark.yml

@@ -6,7 +6,8 @@ on:
 
 jobs:
   benchmark:
-    runs-on: ubuntu-latest
+    if: ${{ github.repository_owner == 'deepmodeling' }}
+    runs-on: codspeed-macro
     steps:
     - uses: actions/checkout@v4
     - name: Set up Python