Variance reduction 'secondary biasing': description of geometry is misleading

[bwheelz36]

As discussed in the forums, specification of the ROI should be done in world coordinates rather than component coordinates - the documentation should reflect this, perhaps using the .. warning directive to reflect that this is a temporary state of affairs...

I also note that if your ROI is at worldZ=0, VRT does not work, but if it is at 1 mm (or I assume any non zero value) then it does work...

[jiperl]

@JoseRamosMendez Hoping you can look into this