it uses HAVE_CUDA, HAVE_NVCUVID, etc. macroses

Looks like we don't actually use them; when creating precompiled headers
with Visual C++, a dummy .cpp is created at build time.

This allows us to forget about the GTEST_(CREATE|LINKED_AS)_SHARED_LIBRARY
macros and to get rid of the dependency on cvconfig.h.

WITH_WINRT -> ENABLE_WINRT_MODE;
Some temporary char* replaced with Ptr<char>;
Build fix for regular WIN32;
Windows Platform SDK and MSVC search added to cmake;
Warinig fixes.

used new CV_PERF_TEST_MAIN_WITH_IMPLS mechanism

Also, shrunk the corresponding column and the test name column.

That way, core/internal.hpp doesn't have to depend on cvconfig.h, which we
don't ship.

glibc defines major and minor macros
which conflict with  gpu::DeviceInfo methods with the same name

Now it can determine, without looking at the file name, both the module
name and the configuration name (the latter with a little help from the
configuration file).

Turns out, you can't use preprocessor directives inside macro arguments.
Who'd have thought?

* In comparison column headers, switched the order of labels, so that
  it's "to" vs "from".

* When a test was present, but not run successfully, put its status in
  the corresponding cell instead of coloring it gray.

Now it allows choosing between arbitrary implementation variants.

Namely, normalize their names to a common convention and remove useless text
from their values.