Using the same blueprint as in the RGBA_MUL/RGB_MUL pull requests, this one adds, the much simpler to create, SIMD versions of the ADD and MUL special blend modes for blitting in SSE2 and AVX2 flavours.

If you found the SIMD code tricky to understand in the other two PRs this one is a lot more approachable - especially for the RGBA flavours as the 'active' code that actually blends the pixels from one surface with another is just one instruction, helpfully named exactly the same thing as the blend mode. See:

mm_dst = _mm_adds_epu8(mm_dst, mm_src);

which as intel describes it does:

Add packed unsigned 8-bit integers in a and b using saturation, and store the results in dst.

Saturation meaning the values for each pixel are clamped between 0 and 255. In SSE2 we have 128 bit registers - so with four channel values between 0 and 255 in each pixel (red, green, blue and alpha) we can see that 128 divided by 4, then divided by 8 bits per channel, we get four pixels worth of data in each register and can thus speed through our adding operation pretty swiftly.

Everything else is just setup to get us to that single line that does the blend.

AVX2 version works the same except we can do 8 pixels at a time due to the larger registers.

There is one more of these PRs, after this one, for the remaining special effect blend modes MIN & MAX.

N.B.
I expect there is some future improvement work to be done on the standard alpha blend and the pre-multiplied blend, but I'd like to see how these special effect modes work out in user land first - plus I would need some time to work on those improvements when work is less busy :)