Reject meshes with vertices whose unaggregated weights sum to more than 4

[io7m]

If a given vertex has large weight values for a large number (> 4) of bones, then the jcalcium compiler is going to pick the largest four and ignore the rest, even though by definition the contributions from the other bones are not insignificant. This could be surprising as the mesh deformations produced at run-time by jcalcium are clearly not going to match whatever came out of an authoring program such as Blender.

This sort of thing should be caught as early as possible, and so compilation should fail when this situation is detected.

[io7m]

Blender seems to be doing some sort of normalization internally. Need to find out exactly what!