Windows: add guards for some methods that don't work/exist on windows

[Tom-Willemsen]

This PR contains some minimal changes to allow a FastCS driver to start on Windows.

• loop.add_signal_handler() is not available on windows. On windows, signals are always delivered to the main thread of an application. Ctrl-c does correctly stop a FastCS driver in non-interactive mode on windows without this handler.
• StdoutProxy didn't work on windows under pytest - I believe it is conflicting with pytest's stdout capturing. Just use sys.stdout for that case.

FastCS' own unit tests don't pass on Windows even with this change - that would require a larger PR.

[codecov]

Codecov Report

✅ All modified and coverable lines are covered by tests.
✅ Project coverage is 90.84%. Comparing base (c86f760) to head (9ddca41).

Additional details and impacted files

@@            Coverage Diff             @@
##             main     #295      +/-   ##
==========================================
+ Coverage   90.82%   90.84%   +0.01%     
==========================================
  Files          70       70              
  Lines        2486     2490       +4     
==========================================
+ Hits         2258     2262       +4     
  Misses        228      228              

☔ View full report in Codecov by Sentry.
📢 Have feedback on the report? Share it here.

🚀 New features to boost your workflow:

• ❄️ Test Analytics: Detect flaky tests, report on failures, and find test suite problems.

[GDYendell]

Could this be moved to a fixture in the tests to keep the production code unchanged? I think something like this would do it

@pytest.fixture(scope="session", autouse=True)
def patch_stdoutproxy_on_windows(mocker: MockerFixture):
    if os.name == "nt":
        import fastcs.logging._logging as logging_mod

        mocker.patch.object(
            logging_mod,
            "StdoutProxy",
            return_value=sys.stdout,
        )

[Tom-Willemsen]

I think the issues with this are:

• Currently this code path is hit when tests are being collected (i.e. before any fixtures get to run) as it's done at import time. An adapted version does work in a pytest_sessionstart hook, but that seems a bit messy to me.
• This would be needed in every downstream repo that tests on windows - I guess we could export a function that does this from FastCS for use by downstream tests in their pytest_sessionstart hook?

Let me know what you prefer?