Use test tools from win-net-test

[nigriMSFT]

Remove the xdpfnmp, xdpfnlwf and pkthlp test components and consume their equivalents from the win-net-test project.

Resolves #422

[nibanks]

This seems painful to keep in-sync with the dependency. Is there no better way to have this somewhere more easily found?

[nigriMSFT]

I 100% agree. This is the pattern used for the eBPF dependency though (and the pattern used by the wtc project's consumption of fnmp), so its no worse off than how existing dependencies are handled.

[nibanks]

Alright. We can keep it as is for now. But I'm not a big fan. I'd prefer some top-level json file or something that lists the dependencies and version numbers. We can follow up later though.

[nigriMSFT]

Hmm the "Downlevel Functional Tests" are running a downlevel version of the test, but with the uplevel version of the test tool installation. This appears to cause an issue, since the test logic is hardcoded to expect the test adapter to have a certain InterfaceDescription/ifdesc (XDPFNMP) and the test adapters are now installed with a difference ifdesc (FNMP).

[nibanks]

Hmm the "Downlevel Functional Tests" are running a downlevel version of the test, but with the uplevel version of the test tool installation. This appears to cause an issue, since the test logic is hardcoded to expect the test adapter to have a certain InterfaceDescription/ifdesc (XDPFNMP) and the test adapters are now installed with a difference ifdesc (FNMP).

Hmmm. How hard would it be to update the downlevel tests to allow for both? We then have to push a change to downlevel to fix.

[nigriMSFT]

I might be able to change the ifdesc with a bit of a hack. Let me try that first

[nigriMSFT]

Darn the hack in mind doesn't seem to work.

I guess I can open a PR on the release 1.0 branch. The change should be pretty small and scoped to test code only.

[nigriMSFT]

Hmm the "Downlevel Functional Tests" are running a downlevel version of the test, but with the uplevel version of the test tool installation. This appears to cause an issue, since the test logic is hardcoded to expect the test adapter to have a certain InterfaceDescription/ifdesc (XDPFNMP) and the test adapters are now installed with a difference ifdesc (FNMP).

It is actually not as simple as a difference in ifdesc. The mp device name has also churned and the downlevel test code is compiled with xdpfnmpapi.lib which uses the old device name to communicate with the driver (xdpfnmp vs. fnmp). :(

Maybe we should just merge this whole PR into downlevel as well?
Or think about if the relevant test tools can be archived with the test dll? That route seems more confusing and error prone though.