Float32 emulation

[dy]

Now values/operations converted from GLSL get increased precision, since JS performs float64 math, whereas OpenGL does float32. So in order to reproduce errors it may be necessary to convert operations to float32, for that purpose to-float32 may come handy.