Assorted fixes for MSVC 2017

[miscco]

There are some issues with MSVC 2017.

Namely it struggles mightily with common_reference_with and weakly_incrementable

I have no idea where things go wrong. Im my analysis of the failing tests each atom of common_reference_with was satisfied on its own but put together it would fail.

Given that 2017 is not receiving any updates I went ahead and simply ddisabled whatever is failing

[miscco]

With this I am getting:

C:\src\cccl\libcudacxx\build>lit test/std/ -sv
lit: C:\src\cccl\libcudacxx\test\lit.cfg:45: note: Using configuration variant: libcudacxx
lit: C:\src\cccl\libcudacxx\utils\libcudacxx\test\config.py:364: note: inferred use_system_cxx_lib as: None
lit: C:\src\cccl\libcudacxx\utils\libcudacxx\test\config.py:249: note: detected cxx.type as: nvcc
lit: C:\src\cccl\libcudacxx\utils\libcudacxx\test\config.py:251: note: detected cxx.version as: (12, 2, 91)
lit: C:\src\cccl\libcudacxx\utils\libcudacxx\test\config.py:253: note: detected cxx.default_dialect as: c++14
lit: C:\src\cccl\libcudacxx\utils\libcudacxx\test\config.py:255: note: detected cxx.is_nvrtc as: False
lit: C:\src\cccl\libcudacxx\utils\libcudacxx\test\config.py:283: note: detected host_cxx.type as: msvc
lit: C:\src\cccl\libcudacxx\utils\libcudacxx\test\config.py:285: note: detected host_cxx.version as: (19, 16, 27050)
lit: C:\src\cccl\libcudacxx\utils\libcudacxx\test\config.py:287: note: detected host_cxx.default_dialect as: c++14
lit: C:\src\cccl\libcudacxx\utils\libcudacxx\test\config.py:289: note: detected host_cxx.is_nvrtc as: False
lit: C:\src\cccl\libcudacxx\utils\libcudacxx\test\config.py:391: note: inferred use_clang_verify as: False
lit: C:\src\cccl\libcudacxx\utils\libcudacxx\test\config.py:667: note: inferred language dialect as: c++17
lit: C:\src\cccl\libcudacxx\utils\libcudacxx\test\target_info.py:67: warning: The locale Czech_Czech Republic.1250 is not supported by your platform. Some tests will be unsupported.
lit: C:\src\cccl\libcudacxx\utils\libcudacxx\test\config.py:484: note: inferred long_tests as: True
lit: C:\src\cccl\libcudacxx\utils\libcudacxx\test\config.py:160: note: Using compiler: C:/Program Files/NVIDIA GPU Computing Toolkit/CUDA/v12.2/bin/nvcc.exe
lit: C:\src\cccl\libcudacxx\utils\libcudacxx\test\config.py:161: note: Using flags: ['-v', '-ftemplate-depth=270']
lit: C:\src\cccl\libcudacxx\utils\libcudacxx\test\config.py:165: note: Using compile flags: ['-std=c++17', '-IC:/src/cccl/libcudacxx\\include', '-Xcompiler', '/GR-', '-D_SILENCE_CXX20_CISO646_REMOVED_WARNING', '-D_LIBCUDACXX_NO_RTTI', '-IC:/src/cccl/libcudacxx\\test/support', '-D_CRT_SECURE_NO_WARNINGS', '-DNOMINMAX', '-Xcompiler', '/bigobj', '-include', 'C:/src/cccl/libcudacxx/test/force_include.h', '-IC:/src/cccl/libcudacxx/include', '--extended-lambda']
lit: C:\src\cccl\libcudacxx\utils\libcudacxx\test\config.py:168: note: Using warnings: ['-Xcudafe', '--display_error_number', '-Werror=all-warnings', '-Xcompiler', '/W4', '-Xcompiler', '/WX', '-Xcompiler', '-wd4100', '-Xcompiler', '-wd4127', '-Xcompiler', '-wd4180', '-Xcompiler', '-wd4309', '-Xcompiler', '-wd4996']
lit: C:\src\cccl\libcudacxx\utils\libcudacxx\test\config.py:169: note: Using link flags: ['-lmsvcrt']
lit: C:\src\cccl\libcudacxx\utils\libcudacxx\test\config.py:171: note: Using available_features: ['libcpp-no-rtti', 'msvc', 'libcpp-no-exceptions', 'windows', 'nvcc-12.2', 'nvcc-12', 'long_tests', 'locale.en_US.UTF-8', 'locale.fr_CA.ISO8859-1', 'locale.fr_FR.UTF-8', 'msvc-19.16', 'pre-sm-60', 'c++filesystem-disabled', 'msvc-19', 'nvcc', 'libc++', 'pre-sm-32', 'nvcc-12.2.91', 'LIBCUDACXX-WINDOWS-FIXME', 'pre-sm-70', 'dylib-has-no-filesystem', 'win32', 'c++17', 'pre-sm-90', 'no-aligned-allocation', 'pre-sm-80', 'locale.ru_RU.UTF-8', 'locale.zh_CN.UTF-8']
lit: C:\src\cccl\libcudacxx\utils\libcudacxx\test\config.py:177: note: Adding environment variables: {}
-- Testing: 1558 tests, 16 workers --
Testing:  0.. 10.. 20.. 30.. 40.. 50.. 60.. 70.. 80.. 90..

Testing Time: 1878.59s
  Unsupported      :  260
  Passed           : 1279
  Expectedly Failed:   19

1 warning(s) in tests

[griwes]

LGTM, but merge after #350