Should `SetBlendConstant` take float,float,float,float instead of WGPUColor?

[kainino0x]

WGPUColor is set up to use double because it can represent all of the possible render pass clear values - it's used as a union of f32, i32, and u32.

However, AFAIK, blending, and therefore wgpuRenderPassEncoderSetBlendConstant, never applies to sint/uint type textures, so should we just take four 32-bit floats instead? Slightly more efficient.

[Kangz]

I checked that all APIs do take 32bit floats.