Investigate SetClientSite(null) in AxHost.DisposeAxControl

[lonitra]

While adding tests for AxSystemMonitor in #10675 , it was found that we get an HRESULT error while attempting to dispose of the control. There is an odd call to IOleObject.SetClientSite(null) in AxHost.DisposeAxControl that gives back HRESULT.E_POINTER when attempting to dispose an ActiveX control. We have always been passing null to IOleObject.SetClientSite since Framework so it is unclear whether or not this method has ever succeeded. The docs seems to indicate that it is ok to pass null pointer so we should investigate what is happening in the windows code and determine next steps to resolve this.

Repro (could also use other active x controls):

        using Form form = new();
        using AxSystemMonitor.AxSystemMonitor control = new();
        ((ISupportInitialize)control).BeginInit();
        form.Controls.Add(_control);
        ((ISupportInitialize)_control).EndInit();

Once fixed, we should also update our tests that test ActiveX controls to dispose of them properly.

[JeremyKuhne]

The ATL host CAxFrameWindow is setting null when tearing down (in atlhost.h in atlmfc sources which can be found in the VS folder). It doesn't check the result.

[JeremyKuhne]

Sysmon always returns E_POINTER (internal - ioleobj.cpp). COleControl in MFC actually expects null to be sent (COleControl::XOleObject::SetClientSite).

I suspect that this is a "bug" in Sysmon. I would say that this is also a cautionary tale for us in regard to being tolerant of return values.