previously assuming that voxel axes equaled display space axes. Also,
bugfix in transform.scaleOffsetXform

vertex shader.

2. Bounds test used in gl21/[tensor/volume]_frag moved to separate file.

3. Transformation matrix offset by 0.5, in Display class, removed, as
I've decided that it should be unnecessary.

called. Fixed. GL21/glvolume no longer using PyOpenGL.vbo module, as it
is rubbish.

calculated in one call.

numpy.dot). And a few tweaks elsewhere

between each of the three spaces (voxel, display, and world)

matrix), LightBoxPanel (not passing displayCtx to LightBoxCanvas),
fslview_parseargs (using imageList.bounds instead of displayCtx.bounds),
ImageList (didn't remove bounds-related code), ImageDisplay (attempting
to offset transformation matrices for affine, when only supposed to do
it for id/pixdim)

to use transform module instead of image.voxToWorld/worldToVox methods

coordinate systemss, with the ultimate aim of having the location panel
always reporting image world coordinates, regardless of the way in which
images are being displayed.