ComputedNodeTarget improvements

[ickshonpe]

Objective

Couple of things need improving with ComputedNodeTarget:

• There aren't aren't any doc comments for its fields and methods.
• The camera field should be optional with absent cameras represented by None instead of Entity::PLACEHOLDER.
• The inverse_scale_factor field should be removed from ComputedNode, ComputedNodeTarget should be the single source of truth for scale factor. ComputedNodeTarget is updated in Prepare and then ComputedNode isn't updated until Layout. So it could cause errors if any systems use ComputedNode's inverse_scale_factor after update_ui_context_system in Prepare but before ui_layout_system in Layout.
• update_ui_context_system is misnamed, intially ComputedNodeTarget was called UiContext and this system should have been renamed to match the new component name.
• All ComputedNodes should be marked as changed on changes to their ComputedNodeTarget to make certain everything is recomputed.

Solution

• Added doc comments to ComputedNodeTarget.
• Remove ComputedNode's inverse_scale_factor field.
• Make the camera field of ComputedNodeTarget optional.
• Rename update_ui_context_system to compute_node_targets_system.
• Add a query for ComputedNodeTargettotext_systemand remove theinverse_scale_factorparameter fromqueue_text` function.
• UI extraction functions that require the scale factor for responsive coordinate resolution now retrieve it from ComputedNodeTarget.
• In ui_layout_system call set_changed on every descendent's ComputedNode if a root node's ComputedNodeTarget is changed.

Testing

If you change your desktop's scale factor with any of the UI examples running you can that the layout and text is recomputed correctly in response.

The multiple_windows and split_screen demonstrate camera target propagation still works correctly.