fix: unbreak NewAPIErrorWithRaw and NewAPIError-related tests

[rainest]

In retrospect from #237, we actually do want to break the exported NewAPIError function. AFAICT we use this in one place only, and that place is a test within KIC that we can change trivially (searched through deck and don't see any use there).

Introducing the separate one apparently broke all sorts of other things in go-kong's internal test expectations and apparently this repo is not configured to block auto-merge on that 😭

https://github.com/Kong/go-kong/actions/runs/3962834113/jobs/6790000801 suggests that the skipped test-passed actually satisfies the branch protection requirements, somehow.

[codecov-commenter]

Codecov Report

Base: 52.84% // Head: 52.87% // Increases project coverage by +0.03% 🎉

Coverage data is based on head (24c5862) compared to base (21ef700).
Patch coverage: 61.90% of modified lines in pull request are covered.

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #267      +/-   ##
==========================================
+ Coverage   52.84%   52.87%   +0.03%     
==========================================
  Files          51       51              
  Lines        4627     4647      +20     
==========================================
+ Hits         2445     2457      +12     
- Misses       1659     1665       +6     
- Partials      523      525       +2     

Flag	Coverage Δ
2.1.4	35.82% <61.90%> (+0.10%)	⬆️
2.2.1.3	45.92% <61.90%> (+0.06%)	⬆️
2.2.2	35.82% <61.90%> (+0.10%)	⬆️
2.3.3	37.55% <61.90%> (+0.09%)	⬆️
2.3.3.4	45.92% <61.90%> (+0.06%)	⬆️
2.4.0	37.63% <61.90%> (+0.09%)	⬆️
2.4.1.3	45.92% <61.90%> (+0.06%)	⬆️
2.5.1	37.63% <61.90%> (+0.09%)	⬆️
2.5.1.2	45.92% <61.90%> (+0.06%)	⬆️
2.6.0	37.63% <61.90%> (+0.09%)	⬆️
2.6.0.2	45.92% <61.90%> (+0.06%)	⬆️
2.7.0	37.63% <61.90%> (+0.09%)	⬆️
2.7.0.0	47.60% <61.90%> (+0.05%)	⬆️
2.8.0	37.63% <61.90%> (+0.09%)	⬆️
2.8.2.2	47.60% <61.90%> (+0.05%)	⬆️
3.0.1	37.03% <61.90%> (+0.09%)	⬆️
3.0.1.0	48.09% <61.90%> (+0.05%)	⬆️
3.1.0	38.67% <61.90%> (+0.09%)	⬆️
3.1.0.0	49.73% <61.90%> (+0.04%)	⬆️
community	39.27% <61.90%> (+0.08%)	⬆️
enterprise	50.33% <61.90%> (+0.04%)	⬆️
integration	52.87% <61.90%> (+0.03%)	⬆️

Flags with carried forward coverage won't be shown. Click here to find out more.

Impacted Files	Coverage Δ
kong/response.go	79.31% <53.84%> (-20.69%)	⬇️
kong/error.go	89.65% <75.00%> (-5.59%)	⬇️

Help us with your feedback. Take ten seconds to tell us how you rate us. Have a feature suggestion? Share it here.

☔ View full report at Codecov.
📢 Do you have feedback about the report comment? Let us know in this issue.

[pmalek]

I think I'd just try to address the failing test rather than doing that.

Or perhaps change the API to accept options like WithBody([]byte)? Passing nil in places where an argument is "not really necessary" feels like a design flaw to me.

[rainest]

Yah, sorry, my bad, was overthinking it and mixed up the tests.