Version 1.1.1

Feature Changes in this release:
1.Add the AVX and FMA3 instruction set support and remove sse2.
2.Deeply optimize the code, so the speed of this single filter (Actual testing method is below) is more than 15 times comparing to version 1.0.1, 25 times to mvf.PlaneCompare.
Because it is so fast, we use src = core.std.BlankClip(width=1920, height=1080, length=1, format=vs.GRAY8)*100000 as source to avoid being encumbered by the source filter, and compared src and src[1:]. At my 5950x CPU, the fps of version 1.1.1 is 11402, while version 1.0.1 is 752, mvf.PlaneCompare is 424.
3. Add propname parameter.

Version 1.0.1

Change filer name from 'psnr' to 'PSNR'
Add Version-getting method.