chore(performance-issues): Create experimental N+1 API Call detector

[leeandher]

For context, I'm auditing the current performance issue detectors and will likely be making some changes to them. Risky changes like touching fingerprinting should probably run silently beforehand to ensure we catch false positives or fingerprint explosion (see guidance).

In this PR, I'm creating a structure to make these experiments easier to review and colocated since I'll be working on different detectors in parallel.

Going forward, my workflow will probably be the following:

1. Copy the detector and test files into /experiments and set up a dummy group type. This copy will be identical to the current detector. (1st PR, This one)
2. Apply experimental changes in a follow up PR (2nd PR, making it easier to review)
3. Add the options via sentry createissueflag (docs) to run in prod (3rd PR, in options-automator)
4. If needed, follow up with a fix/change to the experimental detector. Repeat as needed
5. When happy with the output, apply changes to the real detector and delete experiment files.

The reason I'm being so formal about this all is because I'll be parallelizing these experiments so I gotta keep this straight for myself as well 👍

[leeandher]

Copy of the current test file with small edits for the experimental group type

[leeandher]

Copy of the current detector file with small edits for the experimental group type

[leeandher]

fyi: Format on save did this

[codecov]

Codecov Report

Attention: Patch coverage is 96.78571% with 9 lines in your changes missing coverage. Please review.

✅ All tests successful. No failed tests found.

Files with missing lines	Patch %	Lines
...ctors/experiments/n_plus_one_api_calls_detector.py	93.87%	9 Missing ⚠️

Additional details and impacted files

@@           Coverage Diff            @@
##           master   #90070    +/-   ##
========================================
  Coverage   87.72%   87.72%            
========================================
  Files       10247    10248     +1     
  Lines      578341   578601   +260     
  Branches    22693    22691     -2     
========================================
+ Hits       507347   507606   +259     
- Misses      70551    70552     +1     
  Partials      443      443            

[roggenkemper]

do we want to create new DetectorType for each of the experiments? or is that just adding bloat that we will need to delete later once we replace the original ?

[leeandher]

Good catch! we don't need to for this case because the DetectorType is only used for settings (the ones in project > performance > [Detector Name]). It'd make sense to make a new one if we wanted to trial any new values or defaults, but we don't have to in this case.

[roggenkemper]

thanks for putting the effort into detailing how we are going to trial these experiments

[asottile-sentry]

reverting as this is breaking CI

[getsentry-bot]

PR reverted: 657d08b