fix(bench): Remove unnecessary Python version matrix from benchmark workflow

The benchmark job was incorrectly configured to run across multiple Python versions using a matrix strategy, which is incompatible with CodSpeedHQ's benchmarking service. This change removes the matrix configuration and standardizes the benchmark execution to a single environment, ensuring proper benchmark collection and reporting.

This fix resolves the issue where benchmarks would fail due to CodSpeedHQ not supporting multiple runs of the same benchmark across different Python versions.

Author: liblaf

template/.github/workflows/bench.yaml

@@ -44,6 +44,7 @@ jobs:
             echo 'has-benches=false' >> "$GITHUB_OUTPUT"
           fi
 
+  # CodSpeedHQ does not support the same benchmark multiple times
   bench:
     name: Bench
     needs:
@@ -55,8 +56,6 @@ jobs:
         uses: actions/checkout@v5
       - name: Setup Python
         uses: liblaf/actions/setup-python@v1
-        with:
-          python-version: ${{ matrix.python-version }}
       - name: Install Mise
         uses: jdx/mise-action@v3
       - name: Bench
@@ -66,7 +65,3 @@ jobs:
           mode: instrumentation
         env:
           FORCE_COLOR: 1
-    strategy:
-      matrix:
-        python-version: ${{ fromJson(needs.metadata.outputs.python-versions) }}
-      fail-fast: false