Use self hosted runner to run benchmark (#5695)

Part of #4537

These are some benchmark results from the `actuated` runner. It is
pretty clear to me that this runner does not have a stable environment
to produce stable benchmark results:
-
https://github.com/open-telemetry/opentelemetry-go/actions/runs/10278899889
-
https://github.com/open-telemetry/opentelemetry-go/actions/runs/10275384692

Thus, I am switching the runner to the self-hosted runner, which uses
bare metal, to run benchmarks.

This is the request to use this type of runner.
open-telemetry/community#2266

The underlying machine is m3.small.x86. It has 8 cores 64 GiB.
https://deploy.equinix.com/product/servers/m3-small/

---

This is an example of using this runner from otel java.
https://github.com/open-telemetry/opentelemetry-java/actions/runs/10277337397/job/28439208528

---

This runner cannot be triggered on a forked repo:
XSAM@faab7b9.
So, it is quite safe if we only let it run on the main branch.

.github/workflows/benchmark.yml

@@ -10,7 +10,7 @@ env:
 jobs:
   benchmark:
     name: Benchmarks
-    runs-on: actuated-arm64-4cpu-16gb
+    runs-on: self-hosted
     steps:
       - uses: actions/checkout@v4
       - uses: actions/setup-go@v5