Don't run benchmarks during CI (#1029)

We don't need to run benchmarks in CI,
especially because it's difficult to get consistent numbers out of it.

We're experimenting with [Bencher](https://bencher.orijtech.com/) for
continuous benchmarking.

.github/workflows/go.yml

@@ -47,6 +47,3 @@ jobs:
 
     - name: Upload coverage to codecov.io
       uses: codecov/codecov-action@v1
-
-    - name: Benchmark
-      run: make bench