Add KHR_TF_2D extension to support 2D Transfer Functions for Volumes

[griffin28]

When a scalar volume value is associated with multiple materials/boundaries a 1D transfer function is unable to render them in isolation. This arises a lot in medical datasets, like the one below, and when trying to isolate different layers in an ICF capsule dataset. There are other limitations to 1D transfer functions that can be found with a quick Google search so I won’t list them here.

APIs like VTK (see VTK Volume Property) currently support 2D transfer functions and knowing what renderers support 2D TFs will be important for volume rendering these types of datasets.

EXAMPLE: Volume rendering head dataset (Data/headsq/quarter) with VTK using OpenGL (vtkOpenGLGPUVolumeRayCastMapper ) with 2D transfer function (linear, gradient) vs. VTK using ANARI (vtkAnariVolumeMapper ) using a 1D transfer function (linear)

[griffin28]

Simple case (Implicit 2D transfer function for scalar data)

1. SpatialField stays the same (single scalar value)
2. Add gradient parameter to scivis volumes
3. Add tf2D parameter to scivis volumes of type Sampler with subtype image2D for explicitly defining the 2D transfer function look-up table with the second dimension representing the gradient magnitude

More general case (Explicit and implicit 2D transfer function for multivariate data)

1. SpatialField supports up to vec2 scalar values
2. Add second opacity parameter to scivis volumes that maps the second scalar value to an opacity multiplier
3. Add tf2D parameter to scivis volumes of type Sampler with subtype image2D for explicitly defining the 2D transfer function look-up table