added unload_current_scene to SceneTree and removed ability for change_scene_to to continue when provided null

[twobitadder]

This is the bigger half of the change mentioned in #63565 - this introduces a check for null on change_scene_to (it provides the same error of ERR_CANT_CREATE as if it were unable to instantiate the PackedScene).

In order to provide the same functionality as that code, unload_current_scene was added which just nukes current_scene. Based on my working through the code, it seems like that was originally the path if change_scene_to was given a null value.

This is my first time interacting with the source code in anything more than a minor capacity, so comments and suggestions are absolutely welcome and encouraged, of course. I wouldn't be surprised if it could use some work to be in line with spec.

Thanks!

[twobitadder]

Case in point - just learned about clang-format. Revised files incoming.

[Diddykonga]

Personally I would prefer the naming remove_current_scene() instead, just because "unloading" sounds like it will remove the Scene from Load Cache, which it does not do, just a matter of semantics though.

[twobitadder]

I would be more inclined to agree if it passed back the scene, so that it would be more in line with behavior from remove_child(). As it stands though, this does at least attempt to memdelete() the object in question. If others agree with the name change though, I'm happy to do it.

[twobitadder]

Superseded by #71105 so i'll go ahead and close!

[YuriSizov]

Yes, sorry for that!

[twobitadder]

No problem at all! It's good to have a solution and it's hard to argue against one from reduz, haha