TravelingActionPotentialNode sets computeShapeBounds on a shape, which seems incorrect

[jbphet]

Noticed the following code in TravelingActionPotentialNode.js during the code review:

    function updateShape() {
      var transformedShape = mvt.modelToViewShape( travelingActionPotential.getShape() );
      transformedShape.computeShapeBounds = computeShapeBounds;
      foreground.setShape( transformedShape );
      background.setShape( transformedShape );
    }

Setting the computeShapeBounds function to be essentially stubbed out is a trick that we sometimes use to make a shape render quickly if we know that it isn't going to be manipulated by the user. However, this is generally done on a node, and not a shape. I'm thinking that the code that sets this method for the transformedShape is not needed, and that the other code in the same file overrides the method for the background and foreground nodes is sufficient.

Assigning to @AshrafSharf for discussion if needed, or just changing the code if he agrees.

[AshrafSharf]

computeShapeBounds function assignment to Shape instance removed

[jbphet]

Looks good, closing.