fix(workflow): update benchmark server start and stop commands

Update the command to start `ledger serve` without using `systemctl`.
Add a command to kill the server process after benchmarking.

.github/workflows/benchmark.yml

@@ -18,7 +18,7 @@ jobs:
           fetch-depth: 0
       - run: go build -o /tmp/ledger ./
       - run: echo "running actions as ${USER}"
-      - run: sudo systemctl restart ledger
+      - run: /tmp/ledger serve --postgres-uri=postgres://formance:[email protected]/ledger --postgres-conn-max-idle-time=120s --postgres-max-open-conns=500 --postgres-max-idle-conns=100 &
       - run: >
           earthly
           --allow-privileged
@@ -29,6 +29,7 @@ jobs:
           --allow-privileged
           ${{ contains(github.event.pull_request.labels.*.name, 'no-cache') && '--no-cache' || '' }}
           ./test/performance+generate-graphs
+      - run: kill -9 $(ps aux | grep "ledger serve"| grep -v "grep" | awk '{print $2}')
       - uses: actions/upload-artifact@v4
         with:
           name: graphs