Implement remaining model data as Float32Collections.

[dawsonc623]

In order to efficiently manage indices, the Uint16Collection type was created. This was used as a model to store hierarchical data in a flat array, and thus prevented the need to flatten the data each time it was requested as a flat array (for rendering purposes).

The implementation of Uint16Collection only addressed the simple cases necessary for indices. A Float32Collection implementation should tackle the more complex cases preventing the flatten process from being removed from the requests for the rest of the data as flat arrays. A huge performance boost will come from this.