Fix T44707: cycles border render regression

Cycles: check for F16C support with __cpuid, as we do for BMI and BMI2

Cycles: And yet another compilation fix after half-float commit for clang.

Suggested by Brecht, tested with gcc > 4.4 and Clang

Cycles: fix compilation on 32-bit Windows for half-floats

Reported by IRC user HG1.

Cycles: Add -mf16c for previous commit for Scons

Thanks to Dingto for noticing!

Cycles: Use native float->half conversion instructions for Haswell CPUs.

This makes OCIO viewport color correction a little bit faster (about -0.5s for 100 samples)
Also set max half float value to 65504.0 to conform with IEEE 754.

Cycles: Use curve approximation for blackbody instead of lookup table

Now we calculate color in range 800..12000 using an approximation a/x+bx+c for R and G and ((at + b)t + c)t + d) for B.
Max absolute error for RGB for non-lut function is less than 0.0001, which is enough to get the same 8 bit/channel color as for OSL with a noticeable performance difference.
However there is a slight visible difference between previous non-OSL implementation because of lookup table interpolation and offset-by-one mistake.
The previous implementation gave black color outside of soft range (t > 12000), now it gives the same color as for 12000.

Also blackbody node without input connected is being converted to value input at shader compile time.

Reviewers: dingto, sergey

Reviewed By: dingto

Subscribers: nutel, brecht, juicyfruit

Differential Revision: https://developer.blender.org/D1280

Cycles: add better specializations for SSE shuffle function and few more wrappers.

Fix T41066: An actual fix for curve intersection on FMA-enabled CPUs

Use native float math functions for MSVC12

`double` surrogates are slow (e.g. pow is 2x slower than powf), and MSVC12
supports fp-math functions from C99.