Degraded performance with Apollo Server and v1

**Describe the bug**

Apollo Server with GraphQL Modules `v1` appears to be twice as slow as it was with `v0`.

**To Reproduce**

I've created a [simple repo](https://github.com/dackland/gql-modules-performance-testing) with three basic servers to demonstrate the performance differences. A load testing script performs as many requests as possible within 10 seconds.

| Server | Median Request Duration | Request Count |
| :---------------: | :---------------------: | :-----------: |
| Apollo Standalone | 236ms | 42 |
| Apollo + GQL Modules v0 | 229ms | 43 |
| Apollo + GQL Modules v1 | 460ms | 22 |

Each server uses the same schema, resolver, and JSON dataset. See the repo for more details and instructions. The test uses a larger dataset (3MB), but we observed a similar 100% latency increase even with smaller payloads.

**Expected behavior**

Using Apollo Server with GraphQL Modules `v1` is similar or better performance to `v0`.

**Environment:**

- OS: macOS
- graphql-modules: `1.4.2`
- NodeJS: `16.3.0`

**Additional context**

Using `createApolloExecutor` (as implemented [here](https://github.com/Urigo/graphql-modules/blob/master/packages/graphql-modules/tests/third-parties.spec.ts#L214-L219)) instead of `createSchemaForApollo` seems to address the performance problem. However, using the custom executor broke field usage and resolver tracing in Apollo Studio.

I noticed [your benchmarks](https://github.com/Urigo/graphql-modules/blob/master/benchmark/basic.case.ts) for Apollo Server are all using `createApolloExecutor`, so it isn't capturing the slowdown that seems to happen when using `createSchemaForApollo`.

The [docs](https://www.graphql-modules.com/docs/get-started) recommend `createSchemaForApollo` for Apollo Server, but the [API Reference](https://www.graphql-modules.com/docs/api) calls both `createSchemaForApollo` and `createApolloExecutor` experimental.

Is there a recommended way of using `v1` with Apollo Server that is performant and works with Apollo Studio? Thanks!