ref(test): Refactor Node integration tests to prevent nock leaks.

[onurtemizkan]

Currently, Node integration tests initialize nock interceptors on server initialization and end them after a certain timeout. This recently created problems with asynchronous event processors, and caused random cases of envelope leaks between tests.

To resolve it, this approach returns http.Server and nock.Scope instances back to the test, to eventually be provided to a helper that either collects envelopes or makes requests. The responsibility of ending those two instances is moved to the collection helpers like getEnvelopeRequest or getAPIResponse.

Tests that do not use those helpers should close the server and nock interceptors manually.
This way, we ensure that we do not lose the context of the server and the nock interceptors between tests.

Needs to be tested on #5512

[AbhiPrasad]

Can we update the README with these details?

IMO not the end of the world for us to have this explicit passing of server and scope. If we think it's too messy, we could also choose to create a class that manages this, and then have the url/server/scope be internal vars to that class. What do you think about that?

If we choose to do the class method, let's first merge this in to unblock Tim, and then come back to it.

[AbhiPrasad]

did we have to add this dep?

[onurtemizkan]

TS falls back to the global installation of nock, which seems to have a slight difference between what we use in node-integration-tests. Had to add this (the same version with node-integration-tests) to resolve it.

[AbhiPrasad]

Now that there are three options, can we pass this in as an options obj (2nd arg)?

[timfish]

👍

[onurtemizkan]

If we think it's too messy, we could also choose to create a class that manages this, and then have the url/server/scope be internal vars to that class.

Yes, I think it would make it much cleaner and easier/safer to use. 👍

[AbhiPrasad]

Awesome, thanks for taking a look at this @onurtemizkan!