test_diagnostic_information is deficient

Description of issue

While runing tox on my system, a Mac running macOS Mojave 10.14.6 (18G95), I see that the unit test test_diagnostic_information fails.

bin/nifake/nifake/unit_tests/test_session.py::TestSession::test_buffer_converter PASSED
bin/nifake/nifake/unit_tests/test_session.py::test_diagnostic_information FAILED
bin/nifake/nifake/unit_tests/test_session.py::test_dunder_version Version = 1.1.3.dev0
PASSED

=========================================================================================== FAILURES ============================================================================================
__________________________________________________________________________________ test_diagnostic_information __________________________________________________________________________________

    def test_diagnostic_information():
>       info = nifake.print_diagnostic_information()

bin/nifake/nifake/unit_tests/test_session.py:1281: 
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _
bin/nifake/nifake/__init__.py:85: in print_diagnostic_information
    info = get_diagnostic_information()
_ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _ _

    def get_diagnostic_information():
        '''Get diagnostic information about the system state that is suitable for printing or logging
    
        returns: dict
    
        note: Python bitness may be incorrect when running in a virtual environment
        '''
        import os
        import pkg_resources
        import platform
        import struct
        import sys
    
        def is_python_64bit():
            return (struct.calcsize("P") == 8)
    
        def is_os_64bit():
            return platform.machine().endswith('64')
    
        def is_venv():
            return 'VIRTUAL_ENV' in os.environ
    
        info = {}
        info['os'] = {}
        info['python'] = {}
        info['driver'] = {}
        info['module'] = {}
        if platform.system() == 'Windows':
            try:
                import winreg as winreg
            except ImportError:
                import _winreg as winreg
    
            os_name = 'Windows'
            try:
                driver_version_key = winreg.OpenKey(winreg.HKEY_LOCAL_MACHINE, r"SOFTWARE\National Instruments\NI-FAKE\CurrentVersion")
                driver_version = winreg.QueryValueEx(driver_version_key, "Version")[0]
            except WindowsError:
                driver_version = 'Unknown'
        elif platform.system() == 'Linux':
            os_name = 'Linux'
            driver_version = 'Unknown'
        else:
>           raise SystemError('Unsupported platform: {}'.format(platform.system()))
E           SystemError: Unsupported platform: Darwin

bin/nifake/nifake/__init__.py:60: SystemError
================================================================================= 1 failed, 127 passed in 9.13s =================================================================================

This reveals that this specific unit test is not mocking the code used to extract information. Specifically platform.system and winreg. This means that when the test runs on a Windows machine, one codepath is taken. When run on Linux a different one, and when run on Mac a different failing path is taken. This also affects code coverage.

The correct thing to do would be to mock those things and to improve the test for the Windows, Linux, other OS cases. This would verify we go through all the codepaths as we should.

Steps to reproduce issue

On a Mac terminal, pull master and then run
tox

Start with bin/nifake/nifake/unit_tests/test_session.py::test_diagnostic_information and the get_diagnostic_information implementation in bin/nifake/nifake/init.py. Run the test under tox, then cover the Windows, Linux, and other-OS branches by mocking platform.system and winreg. Done means the diagnostic test is platform-independent and exercises all relevant code paths.