ref(tests): Refactor Node integration tests further.

[onurtemizkan]

Following up: #5579

Apparently, the updates on #5579 were not enough to resolve leaking issues on #5512.

Debugging further, it seems that whenever a manual http request is initiated from a test (or test server) without waiting for / asserting on events or transactions, there are also transactions as side effects which may or may not finish before the test ends. This specifically happens with tests that assert request headers for trace and baggage, because they don't need to wait for the request / scenario to finish.

So it looks like the leakage to the next test doesn't come from the nock interceptor or an unclosed server, it comes from an unflushed Sentry transaction. The other tests using getEnvelopeRequest or getMultipleEnvelopeRequest finish wait for the events to be flushed.

While debugging this, I also made some more refactors in utilities, which I think will make them more concise.

Summary:

• Making sure the events are flushed before tests that use getAPIResponse or runScenario complete.
• Converted http usages in helpers to axios calls for simplicity and debuggability.
• Moved nock initialization out of runServer so we don't need to carry nock.Scope around anymore. The nock instance is either manually initialized, or inside the getMultipleEnvelopeRequest function. The tests that are using manual http calls don't normally need nock anyways.
• Filtering envelopes by their types on nock level, so if there is a need to filter them, we don't lose the expected count. Also, that makes the envelope indices inside tests simpler.
• Switched to nock.cleanAll() and nock.removeInterceptor(mock) to end nock instead of .persist(false).

I have tested this locally on #5512 and it seems to be working well.
🤞

[AbhiPrasad]

@onurtemizkan - what do you think about trying to turn this into a class while we are here doing fixes? #5579 (review). If you think it adds too much complexity, no worries, we table it for later.

Also, happy to pair program on this any time! Just let me know :)

[onurtemizkan]

what do you think about trying to turn this into a class while we are here doing fixes? #5579 (review).

@AbhiPrasad, yes I was planning to include that, but would prefer if we do that on a clean scope after #5512 gets merged, if that's OK? I'm not entirely sure if there is still a potentially fragile test around.

[AbhiPrasad]

yes I was planning to include that, but would prefer if we do that on a clean scope after #5512 gets merged, if that's OK

Sounds like a plan @onurtemizkan, lets do that!

[AbhiPrasad]

Nice - I think this is cleaner and easier to understand as well!

[github-actions]

size-limit report 📦

Path	Size
@sentry/browser - ES5 CDN Bundle (gzipped + minified)	19.41 KB (+0.03% 🔺)
@sentry/browser - ES5 CDN Bundle (minified)	60.06 KB (0%)
@sentry/browser - ES6 CDN Bundle (gzipped + minified)	17.98 KB (+0.02% 🔺)
@sentry/browser - ES6 CDN Bundle (minified)	52.92 KB (0%)
@sentry/browser - Webpack (gzipped + minified)	19.77 KB (0%)
@sentry/browser - Webpack (minified)	64.31 KB (0%)
@sentry/react - Webpack (gzipped + minified)	19.79 KB (0%)
@sentry/nextjs Client - Webpack (gzipped + minified)	44.71 KB (0%)
@sentry/browser + @sentry/tracing - ES5 CDN Bundle (gzipped + minified)	25.92 KB (+0.01% 🔺)
@sentry/browser + @sentry/tracing - ES6 CDN Bundle (gzipped + minified)	24.29 KB (+0.01% 🔺)