Fix Flaky testing framework discovery against actual python process tests

[karrtikr]

The following test with unittest framework often fails on CI both in Azdo as well as in GHA

  1) Unit Tests - unittest - discovery against actual python process
       Re-run failed tests results in the correct number of tests counted:

      AssertionError [ERR_ASSERTION]: This test was written assuming there was 2 tests run that would fail. (iteration 1)
      + expected - actual

      -0
      +2

Example builds:

• https://github.com/microsoft/vscode-python/runs/647842574?check_suite_focus=true
• https://github.com/microsoft/vscode-python/runs/647709884?check_suite_focus=true

Another flaky test under the same suite,

  1) Unit Tests - unittest - discovery against actual python process
       Ensure correct test count for running a set of tests multiple times:

      AssertionError [ERR_ASSERTION]: This test was written assuming there was 2 tests run that would fail. (iteration 1)
      + expected - actual

      -0
      +2

Example build:

• https://github.com/microsoft/vscode-python/runs/655005889?check_suite_focus=true

It maybe the case that the operation itself is flaky (due to use of sockets in unittest framework), and the tests are just reflecting that. If that's the case, these should be fixed by #10119.

I think also recall seeing this failing with other frameworks - pytest, nose before, but I am not sure as they don't fail anymore, so just focusing on the unittest framework for now.

cc @karthiknadig

[eleanorjboyd]

@karthiknadig do you have context for this? Does this problem still exist? The links no longer work and if it is talking about our product of python tests the rewrite made most things different so it might not be applicable.

[karthiknadig]

This was bout the old test code. This can be closed.