CL: Use uniform variables instead of shader_in for kernel inputs
This is just a mechanical replace. The semantics of inputs for CL are a closer match to the semantics of uniforms for graphics. Rather than cross-stage data, it's data that every thread sees uniformly.
This also makes it simpler to deal with images (!5242 (merged)): currently the SPIR-V to NIR converter generates tex and image intrinsics by casting images/samplers to uniform derefs.