QSurfaceFormat does not resolve correcctly

When creating a QSurfaceFormat for a QOpenGLWidget, it is not unreasonable to request a 32-bit depth buffer. Some of our test machines support this and some do not. For those that do not the fallback to the supported value of 24 would be expected. Instead, because of the way QSurfaceFormat resolves to a supported format I loose several of the settings I set in the QSurfaceFormat.

QSurfacceFormat should try reducing the depth buffer first before reducing other values as the depth buffer size has the greatest impact on the requested format not being supported than any other parameter.

Ideally the reduction stage of the QSurfaceFormat would reduce one parameter at a time instead of adding up reductions, so that the final format is closest to the requested format.