Make all fp16 routines work with fp32 as input instead of fp64, since that is what...

[movit] / multiply_effect.frag

vec4 FUNCNAME(vec2 tc) {
        return INPUT(tc) * PREFIX(factor);
}