xlock.cpp: Should we use SRWLOCK instead of CRITICAL_SECTION?

[frederick-vs-ja]

In #2317, MSVC STL dropped support for Windows Vista and make SRWLOCK unconditionally used for mutex. However, there're still some implementation details using CRITICAL_SECTION, all of which can be found in xlock.cpp (via xmtx.hpp and xmtx.cpp).

It seems that switching to SRWLOCK doesn't affect ABI (as support for Vista has been dropped). Should we do this?

[StephanTLavavej]

I've taken a look at the code - it should be possible, as there shouldn't be any mix-and-match concerns from user code (as xlock.cpp's state is static). However, it must be done with great care, as the helper functions appear on the export surface. That is, the _Mtxinit family is exported from msvcp140.dll for our eternal nemesis, /clr:pure. It should be possible to continue to export them (marked as unused and preserved for ABI) while xlock.cpp switches to using SRWLOCK.

[AlexGuteniev]

Make sure that none of these mutexes are used as recursive