RenderCapture pixel values are half the expected value when using RGBA32F textures

RenderCapture relies on GraphicsContext::copyGLFramebufferDataToImage to copy data from the framebuffer into the image. In copyGLFramebufferDataToImage, every pixel value is divided by 1 + the value. This results in values in the range 0 to 0.5 (or 0 to 127), which makes the image darker than expected.

The current implementation of copyGLFramebufferDataToImage appears to attempt to map from some non-linear range of values to a linear range. However, it appears the only way to obtain numbers that are equal both when set in a shader and when read from the QImage, is to remove the division and use the value directly. That is, to map from a linear range to another linear range.