Calculate confidence interval for benchmark measurements (#6950)

.github/workflows/osrm-backend.yml

@@ -700,15 +700,6 @@ jobs:
           mkdir -p $HOME/.ccache
           ccache --zero-stats
           ccache --max-size=256M
-      - name: Build PR Branch
-        run: |
-          mkdir -p pr/build
-          cd pr/build
-          cmake -DENABLE_CONAN=ON -DCMAKE_BUILD_TYPE=Release ..
-          make -j$(nproc) 
-          make -j$(nproc) benchmarks
-          cd ..
-          make -C test/data 
       - name: Checkout Base Branch
         uses: actions/checkout@v4
         with:
@@ -723,9 +714,23 @@ jobs:
           make -j$(nproc) benchmarks
           cd ..
           make -C test/data 
-      - name: Run Benchmarks
+      - name: Build PR Branch
         run: |
-          ./pr/scripts/ci/run_benchmarks.sh base pr
+          mkdir -p pr/build
+          cd pr/build
+          cmake -DENABLE_CONAN=ON -DCMAKE_BUILD_TYPE=Release ..
+          make -j$(nproc) 
+          make -j$(nproc) benchmarks
+          cd ..
+          make -C test/data 
+      - name: Run PR Benchmarks 
+        run: |
+          ./pr/scripts/ci/run_benchmarks.sh -f $(pwd)/pr -r $(pwd)/pr_results -s $(pwd)/pr -b $(pwd)/pr/build -o ~/data.osm.pbf -g ~/gps_traces.csv
+      - name: Run Base Benchmarks
+        run: |
+          # we intentionally use scripts from PR branch to be able to update them and see results in the same PR
+          ./pr/scripts/ci/run_benchmarks.sh -f $(pwd)/base -r $(pwd)/base_results -s $(pwd)/pr -b $(pwd)/base/build -o ~/data.osm.pbf -g ~/gps_traces.csv
+
       - name: Post Benchmark Results
         run: |
           python3 pr/scripts/ci/post_benchmark_results.py base_results pr_results