add SceneNode deletion convenience methods

[ribbon-otter]

This commit adds convenience method to delete a node and its attached objects, as I needed and people asked about here and here.

The use case is if you want to clear a SceneNode and put everything back from scratch (a simple if slower way of updating an object to reflect game state), or if you wish to stop rendering a node (and thus things attached to it can be garbage collected).

This patch adds three new non-virtual methods to SceneNode that analogous to the other methods but also delete the detached objects:

SceneNode::destroyGraphSegment, analogous to SceneNode::removeAndDestroyAllChildren, except it also destroys all MovableObjects attached to this and to any of the descendants. In other words, all children are destroyed and any objects attached to them and any objects attached to this.

SceneNode::destroyChildGraphSegment, analogous to SceneNode::removeAndDestroyChild, except it also destroys the objects attached to the child.

SceneNode::destroyAllObjects, analogous to SceneNode::detatchAllObjects, except it destroys the objects too. Unlike detatchAllObjects, this function is not virtual. I believe it is not required for it to be, since everything it does could be achieved using external methods. Moreover, I did not want to increase memory footprint of SceneNode. I called it destroyAllObjects rather than detatchAndDestroyAllObjects to keep the name short and because detatching is already considered part of destroying MovableObjects.

The naming scheme of GraphSegment was chosen as removeAndDestroyAllChildrenAndAttached seemed too long. removeAndDestroy in the original methods' names feel redundant to me, but best not changed as to not break the API.

There are tests added to test the basic behaviour of the 3 added methods included in this pull request.

This code deletes MovableObjects only; I have never worked with skeletons or animations, so it may not delete those. I am not sure if deleting animations would even be desired behaviour, as animations might be more analogous to meshes than entities.

Please view the code with some skepticism, as I am unfamiliar with the Ogre code base and thus, it is greater risk that I broke expected behaviour or introduced undefined behaviour. And I am of course, happy to make any changes that are needed.

[ribbon-otter]

is there a command to run the trailing whitespace check locally? I have been using make && make test and it doesn't seem to be catching it.

[paroj]

is there a command to run the trailing whitespace check locally?

git diff --check origin/master

[paroj]

please use empty()and front() as below: https://github.com/OGRECave/ogre/pull/2963/files#diff-8b29158c3b96916d83c3b4bbc1c9ee083fee97bff80ae347e6401c6bc49dd5afR358

[paroj]

The naming scheme of GraphSegment

While GraphSegment technically makes perfect sense, it is not used anywhere else within ogre. Therefore I would rather keep to "Children" and "Objects" for now.

[paroj]

Suggested change

	void destroyGraphSegment();
	void destroyAllChildrenAndObjects();

[paroj]

Suggested change

	void destroyChildGraphSegment(const String& name);
	void destroyChildAndObjects(const String& name);

[paroj]

Suggested change

	SceneNode* pChild = static_cast<SceneNode*>(getChildren()[index]);
	SceneNode* pChild = static_cast<SceneNode*>(removeChild(index));

otherwise pChild might be OOB. This is an issue in the original code as well

[ribbon-otter]

I believe I have made the changes you have requested