Windows test

[ImogenBits]

I don't really have an actual point or good idea here, but I wonder what the correct way of doing the windows test are.

With the way that github actions are set up you can't run linux containers there, so we need to basically support not just algobattle running on windows, but algobattle running windows containers on windows. I don't think that that is ever going to be an actual use case. You can't have only some images be windows so this would only happen if someone decided all their containers are windows.

There also is the problem that debugging them is incredibly annoying, there are no good "base" windows images, the shells running on the windows images are different from the ones running locally, etc. This is now just making us ignore the failing CI checks cause we know they don't really mean anything.

Guaranteeing windows support is very important imo but I don't think the current CI is the way to do that. I don't know what a better way would be, just making a space here for that train of thought to stay.

[Benezivas]

The main goal of the CI checks is to ensure that the code lives up to the standard formulated by the tests.

What I think we should aim to test is the basic codes basis being able to run without issues on all host platforms, but not iterate through every possible base platform of the docker images being built during the tests.

I think we talked briefly about optionally enforcing the platform for docker containers that are built from problem images. This might be a compromise to do the CI checks. I am however far from an expert on the CI on Github regarding Windows and MacOS images. If these checks are currently infeasible and/or unstable we should not do them instead of getting failures for runs that turn out not to be the tool's fault.