In computing the text bounding box, the -1 in the formula was omitted.

NormalizedViewport creates slightly offset lines.

Compute interLineSpacing as an integer and use that instead.

Previously, ascent and descent were distances from baseline.
Descent included the baseline, ascent did not. This was
confusing and undocumented.

This property allows the user to center the string to the anchor point.
By default, the bounding box for the string is computed
using the font metrics which includes ascents and descents.
As descents might not be present in the string, the label will not
be perfectly centered on the anchor point. Setting this property
on will compute the bounding box using the current string rather than
the font metric. This results in perfectly centered labels.
This does not work well for temporal data if the string changes as
it results in text that moves around.

Originally, the full height was computed using n text heights + n interline spaces.

When working on an empty polydata, vtkGlyph3DMapper::GetBounds would
return the bounds of the glyph instead of vtkMath::UninitializedBounds.
This would be confusing because glyph's bounds has nothing to do with
the rest of the scene and discrupts
vtkRenderer::ResetCameraClippingRange. To fix this, a check condition is
added and a new test file is included.

The current dependency relationship in vtk is unclear and misleading.
This MR tries to rewrite them based on header files inclusion of headers
and source files in each module. Corner cases are considered and
modules are sorted in alphabetical order to facilitate future reference.
See details in my gitlab python based script project. In future we can continue
using this script to clean the VTK Dependency easily from
time to time.

it was not taking into account representation

A few tweaks and fixes as detailed below.

Remove the use of a physical scale factor in
computing the camera transforms. Instead use the
distance from the focal point to the camera as a
metric. That distance will be mapped into one
meter in the physical space. This fits the model
of a camera dolly.

This has the advantage of removing the scale factor that
was being added into the World to View matrix. We generally
assume there is not a scale in the world to view matrix and
that was causing some issues with some shaders (such as imposters)

Fix an issue where the scale was being added to the NormalMatrix
The normal matrix should never have a scale.

Adjust the scaling interaction to scale about a meter above the
floor. This is more intuitive than scaling from the floor as an
origin.

Fix a bug in the Oculus code when a buffer submission failed with
a resubmit success error code.

This commit changes the OpenGL2 GPU volume renderer to consider whether the
volume transform preserves orientation when orienting polygons. Before, nothing
was rendered when the transform didn't preserve orientation, presumably due to
backface culling.

The legacy OpenGL volume renderer handles this case.

These mappers are useful for non chemistry applications
so move them to the OpenGL2 directory

comes up in paraview server side rendering in which the client
side has no input

There were lost of std::string operations that resulted
in lots of memory allocs/deallocs. This patch significantly
reduces them. In my testing of vtkCompositePolyDataMapper2
these changes improved the frame rate from 75 fps up to
400 fps.

added a glew check that needed to be ifdef'd out for oepngles

showed up in cellarray test

gradient shadding didn't appear to be working, but only because
the new volume never had a chance to set up its own properties.

Just needed to register it with the factory.
Made the two tests nearly identical in testing.

and minor fixes for tests

command line arg "-EXT number" to control wavelet size
command line arg "-GL" to use OpenGL instead of OSPRay
reports wavelet build time too

Bring over another test and add -GL command lines to ease comparison