Add GraphQL DataFetcher Instrumentation

[hannahchan]

This PR refactors the GraphQL 20.0 instrumentation to introduce the ability to create spans for DataFetchers.

At a high-level;

• Added GraphQLInstrumeterFactory
• Updated GraphQLTelemetryBuilder, GraphQLTelemetry with the new options; createSpansForDataFetchers and createSpanForTrivialDataFetchers
• Updated OpenTelemetryInstrumentation with logic to create the actual spans for DataFetchers
• Updated OpenTelemetryInstrumentationState to correctly track the OpenTelemetry Context for DataFetchers

I was unable to make these changes compatible with previous versions of graphql-java.

[hannahchan]

I'm still yet to verify this but suspect this block is the source of performance regressions in our GraphQL endpoint in our production environment. I'm unable to turn this off as this needs to be sanitised.

We are seeing regressions of around 10-100+ milliseconds added to our GraphQL operations when the OpenTelemetry GraphQL instrumentation is turned on. I suspect traversing the AST is causing the regression with larger GraphQL documents resulting in larger regressions.

I think this block needs to be surrounded with a check on span.isRecording() to at least not impact performance if the span is not sampled.

WDYT?

[hannahchan]

Alternatively we could find a way to off load this onto another thread to do this async as to not block the execution of the GraphQL operation.

[laurit]

If you can then please do verify. If we can reproduce this then we can figure out a better strategy to handle it. If the query is really large then perhaps we can only handle the start of the query and truncate most of it.

[hannahchan]

@laurit

What else do you think I need to do in order to get this over the line? Are there any other concerns that need to be addressed?

[laurit]

@laurit

What else do you think I need to do in order to get this over the line? Are there any other concerns that need to be addressed?

@hannahchan I have create a PR hannahchan#1 for your PR that should get it closer to acceptable

[hannahchan]

@laurit
What else do you think I need to do in order to get this over the line? Are there any other concerns that need to be addressed?

@hannahchan I have create a PR hannahchan#1 for your PR that should get it closer to acceptable

Ah I missed this. I'll take a look and incorporate the changes as soon as I can.

Thanks.

[hannahchan]

@laurit
What else do you think I need to do in order to get this over the line? Are there any other concerns that need to be addressed?

@hannahchan I have create a PR hannahchan#1 for your PR that should get it closer to acceptable

Ah I missed this. I'll take a look and incorporate the changes as soon as I can.

Thanks.

I've merged the above PR into my branch via the command line and updated the unit tests to make sure the data fetcher spans and structure is what we expect it to be under various conditions. I think this is ready to be merged unless there are other things that need to be addressed.