Unless I am misunderstanding something, it seems like you are introducing a lot of unnecessary complexity by using Metal to convert between color spaces. It looks like Core Image can do the job for you?

• -[CIImage initWithCVPixelBuffer:] will build and use the color space from the color space attachments in the image buffer.
• kCIContextWorkingColorSpace can be used to specify the working color space of the CIContext. You would use the same color space as the CIImage (via -[CIImage colorSpace]).
• From here, you can convert the CIImage to any other image format, specifying the destination color space (in your case, sRGB) in the respective conversion functions.
  • Use -[CIImage imageByColorMatchingWorkingSpaceToColorSpace:] to get another CIImage (e.g. to put into a UIImage for display in a UIImageView).
  • Use -[CIContext createCGImage:fromRect:format:colorSpace:] to get a CGImage.
  • Use -[CIContext render:toCVPixelBuffer:bounds:colorSpace:] to render into a CVPixelBuffer.
  • Use -[CIContext render:toIOSurface:bounds:colorSpace:] to render into an IOSurface.
  • Use -[CIContext render:toMTLTexture:commandBuffer:bounds:colorSpace:] to render into a MTLTexture.

(Caveat: I have not personally tested any of these functions except the first one.)