Add first version of a 4x4 AVX512 matrix

Passes all available unit tests
The combinations are a little hard to manage if we want to support all types with vectors as large as possible, e.g. float4x3 is still 3 n128 even though it could in theory be an n256 and an n128
Since 4x4 is the most common it warrants the most attention but perhaps in the future we can do more, it's still quite a bit of work

Fix the incorrect inclusion of arm64_neon.h on unsupported compilers

Technically ARM64 on VS2017 and the experimental Clang might not work but I don't think that is a problem and it'll keep the fix simpler
Also enable the ARM64 and double types for _M_ARM64EC

Add ARM64EC to the list of defines that trigger NEON intrinsics

Fix matrix cast in AVX

vcpkg cmake support (#84)

This is a helper cmake to set the correct variables for include directory and version.
It's meant to be used by vcpkg and is required in order to make the vcpkg port an official port (see https://learn.microsoft.com/en-us/vcpkg/contributing/maintainer-guide#add-cmake-exports-in-an-unofficial--namespace)

Merge branch 'master' of https://github.com/redorav/hlslpp

Fix compile error with template incantation

Use t4 instead of t

Fix uintN swizzle

Merge branch 'master' of https://github.com/redorav/hlslpp