HTTP fault injection returns 502 (bad gateway) for all requests about xk6-disruptor

[pablochacin]

Unit tests did not catch this mistake because the mocked HTTP client never exercises this request. This is the kind of subtlety that integration tests are good at (re. #183)
I am not sure why e2e tests did not catch this.

We already have #183 for this discussion about integration tests, but for me, the solution to "e2e test did not catch it" is not to create integration tests because the e2e tests use an actual HTTP server, so I don't see what value it would add.

I think the problem is much simpler: we in the e2e tests are not checking that we don't introduce errors. We are just checking if we get the error we ask for.

So, we don't need to add more tests, just one test case to the existing e2e tests. For example, with errorRate: 0 (do not inject errors) and checking if we get a 200. I suggest we do it in the context of fixing this issue.

[roobre]

It seems like this was caused by a dumb typo: upstreamReq was cloned and its RequestURI cleared, but req was passed instead of upstreamReq 🤦🏻‍♀️

Unit tests did not catch this mistake because the mocked HTTP client never exercises this request. This is the kind of subtlety that integration tests are good at (re. #183)

I am not sure why e2e tests did not catch this.