Reduce redraws in editor

[wjt]

Fixes #649

[wjt]

Draft for 2 reasons:

1. I've not yet tested this in our game;
2. I've not tested/fixed 3D because I don't use the 3D side of Godot at all.

[ramokz]

Just checking in, has the change worked as expected in your 2D project(s)?

[wjt]

Sort of... I get this warning when switching in the editor from a scene that does not use pcam to a scene which does:

SCRIPT ERROR: Invalid call. Nonexistent function 'get_active_pcam' in base 'Nil'.
          at: _on_update_editor_viewfinder (res://addons/phantom_camera/scripts/panel/viewfinder/viewfinder.gd:552)
          GDScript backtrace (most recent call first):
              [0] _on_update_editor_viewfinder (res://addons/phantom_camera/scripts/panel/viewfinder/viewfinder.gd:552)
              [1] _pcam_priority_override (res://addons/phantom_camera/scripts/phantom_camera_host/phantom_camera_host.gd:1404)
              [2] @priority_override_setter (res://addons/phantom_camera/scripts/phantom_camera/phantom_camera_2d.gd:119)
              [3] _enter_tree (res://addons/phantom_camera/scripts/phantom_camera/phantom_camera_2d.gd:713)

And we also have a strange issue that we have not yet debugged where limits are applied incorrectly when switching camera, but only in (some) exported builds. endlessm/threadbare#2393

I haven't had a chance to test whether these are also reproducible without this stack of patches.

[ramokz]

For some reason I cannot replicate the warning you're seeing, but I am encountering an error in update_dead_zone() with those changes:

Invalid type in function '_set_viewfinder' in base 'Control (viewfinder.gd)'.
The Object-derived class of argument 1 (previously freed) is not a subclass of the expected argument class.

Though it's pretty easy to fix by just adding a check and additional assignment.

Another side effect I'm seeing from this PR is when usingFramed Follow while running a scene and having Show Viewfinder in Play enabled where the yellow dot no longer follows its target's position and now stays at the center of the viewport. Digging a bit further it seems like it's because the viewfinder_visible checker returns false in _update_processing().

framed_follow_dot.mp4

And we also have a strange issue that we have not yet debugged where limits are applied incorrectly when switching camera, but only in (some) exported builds. endlessm/threadbare#2393

The issue in that post has been a problem before in general runtime situations (editor included), but should have been fixed quite a while back at this point. But it's very odd that it's seemingly only happening in some build targets and not consistently... My only hunch is the limit bounds are smaller than the PCam2D's viewport and so the camera doesn't quite know where to position itself.
When testing the web build (from the public page on your repo), I cannot repro the issue in that particular spot, but I do see it happening at the bottom of that level. Not sure if anything has changed in that builds compared to the ones you've been testing on:

recording_15_07_26.mp4