CI/CD jobs can succeed despite test failures.

[rectified95]

Describe the bug

1. There are a few PRs for which the pipeline shows a passing run, even though their logs contain test failures.
   Ex. https://github.com/microsoft/ebpf-for-windows/actions/runs/5672733794/job/15373312646

bindmonitor_tailcall_native_test
-------------------------------------------------------------------------------
D:\a\ebpf-for-windows\ebpf-for-windows\tests\api_test\api_test.cpp([64](https://github.com/microsoft/ebpf-for-windows/actions/runs/5672733794/job/15373312646#step:19:65)4)
...............................................................................

D:\a\ebpf-for-windows\ebpf-for-windows\tests\api_test\api_test.cpp(624): FAILED:
  REQUIRE( perform_bind(&sockets[0], 30000) == 0 )
with expansion:
  -1 == 0

[12:54:44] :: api_test.exe Passed

I've also seen an example of this in other PRs.
Have we added any indirection test invocation and result gathering?

2. Crash dumps are not always picked up by the respective job.
   From the same run as above:

[12:52:40] :: Executing api_test.exe
[12:52:40] :: C:\Program Files\OpenCppCoverage\OpenCppCoverage.exe -q --modules=C:\eBPF --export_type binary:api_test.exe.cov -- api_test.exe
[12:54:31] :: Generation of crash dump succeeded
Error: [12:54:41] :: [error] 

----------------------------------------------------
*** Your application has thrown an unhandled exception. Code: 32[21](https://github.com/microsoft/ebpf-for-windows/actions/runs/5672733794/job/15373312646#step:19:22)[22](https://github.com/microsoft/ebpf-for-windows/actions/runs/5672733794/job/15373312646#step:19:23)5477: EXCEPTION_ACCESS_VIOLATION
----------------------------------------------------
See https://github.com/OpenCppCoverage/OpenCppCoverage/wiki/FAQ#your-application-has-thrown-an-unhandled-exception-code-3221225477-exception_access_violation for additional information.

Error: [12:54:44] :: [error] Your program stop with error code: -2147483645

Then during the step 'Check for crash dumps':
These files don't exist: c:/dumps/x64/Debug/*.dmp

Steps taken to reproduce bug

run the CI

Expected behavior

CI runs never show up as successful when their components fail.

Actual outcome

CI runs sometimes show up as successful when their components fail.

[rectified95]

We missed another regression because of this, will start looking today/tomorrow.
fyi @shankarseal
Failure introduced in https://github.com/microsoft/ebpf-for-windows/actions/runs/6152010592/job/16693947569?pr=2823

[rectified95]

Looks like the test EXE doesn't the exit code properly: