Improve InterpolationTest

[javagl]

"Improve" is a very generic term, so here it is broken down:

• it had a size of 33KB, now it has 18KB
• it contained 10 materials, now it contains 2
• it generated 14 'Infos' and 40 'Hints' (!) when validating, now it generates only 1 'Info' (about the non-power-of-2 image size)
• it had some ... slightly odd, non-looping animations

The last one is the main reason for this update. This was brought up in KhronosGroup/glTF-Sample-Models#391 , with some details described there. The animation data for the linear translations did not loop (i.e. they ended at a different place than where they started), and nearly all values (translations, rotations...) had been ... slightly odd numbers.

EDIT: The following is now obsolete. Apparently, the bug that caused the model to not be rendered in the sample viewer has been fixed - see KhronosGroup/glTF-Sample-Viewer#495

Now... merging this should be uncontroversial, but... ... the updated model does not render in the glTF-Sample-Viewer 😬

I opened KhronosGroup/glTF-Sample-Viewer#495 to track this. Let's see how that plays out...

[javagl]

The scale animations in this sample originally interpolated between (0.5, 0.5, 0.5) and (1.0, 1.0, 1.0). The sample now has been updated so that the scale animations interpolate between (0.0, 0.0, 0.0) and (1.0, 1.0, 1.0).

The main purpose of this is that according to #77, a scale of (0,0,0) (with 'step' interpolation) is sometimes used to "completely hide" a mesh, and some viewer seemed to not handle this properly.