Animations should be grouped by action, not by target node about collada2gltf

[lasalvavida]

This is another good beginner issue. The fix entails adding a function like mergeAnimations to GLTF::Asset, and then calling it from main.cpp during the optimization stage. This function should collect all of the channels and samplers across the animations and either add them all to the first one, removing the rest, or wipe out all existing animation objects and add the channels and samplers to a new one. Animations are not used by reference anywhere else, so this should be a trivial addition.

[donmccurdy]

To add a bit more context for anyone who can look at this, the proposed changes are:

1. Merge all channels and samplers into a single animation, as described above.
2. (Optional) Add support for the COLLADA <animation_clip/> node, to separate channels and samplers for each animation clip into separate glTF animations.

[lasalvavida]

Closed by #102.