feat(instrumentation-graphql): add option to ignore resolver spans

[seemk]

Which problem is this PR solving?

#1739

Short description of the changes

Add the option to ignore resolver spans.

Creating a span for each resolve can cause way too many spans to be created (e.g. 100k elements in the response equals at least 100k spans, most of them will be dropped). This causes huge latency increases - my test query with an in memory resolver went from 1.2s to 8-9s. With ignoreResolveSpans: true the response time went back to ~1.2s.

For reference, the Java implementation of GraphQL instrumentation ignores resolvers by default and there isn't even a possibility to enable creating spans from resolvers.

I think ignoring resolvers should be the default behaviour in the Node.js instrumentation as well, but for backwards compatibility I set the default value to false in this PR. Up for discussion whether to ignore resolvers by default or not.

[codecov]

Codecov Report

Merging #1858 (73e8e6e) into main (c54e9b6) will decrease coverage by 0.03%.
The diff coverage is 100.00%.

❗ Current head 73e8e6e differs from pull request most recent head 72c88ab. Consider uploading reports for the commit 72c88ab to get more accurate results

Additional details and impacted files

@@            Coverage Diff             @@
##             main    #1858      +/-   ##
==========================================
- Coverage   91.50%   91.47%   -0.03%     
==========================================
  Files         145      145              
  Lines        7425     7414      -11     
  Branches     1484     1486       +2     
==========================================
- Hits         6794     6782      -12     
- Misses        631      632       +1     

Files	Coverage Δ
...try-instrumentation-graphql/src/instrumentation.ts	92.81% <100.00%> (+0.63%)	⬆️

... and 4 files with indirect coverage changes

[pichlermarc]

I think this is a good change, thank you for taking care of this. 👍

Given that there's cases where performance is impacted this much, I think it would be reasonable to make ignoreResolveSpans: true the default in a future release.

[seemk]

@pichlermarc Are there any additional approves this PR needs?

[pichlermarc]

@seemk, no other approvals needed. I'll merge this in. Thank you for your contribution!

[satazor]

Just noticed that this PR missed the documentation update on the README. I will make a PR to add it.