Interface ClipBoundingInfo

Unified, host-facing bounding description of a clip in composition space.

All fields describe ONE box: the clip's interactive geometry (width × height, always non-negative — flips are visible through style.getScale() signs, not here). Visual extent beyond the geometry (e.g. a shape's centered stroke and drop shadow) is reported separately through contentPadding and is NOT included in width/height/center, so toggling a shadow never moves selection handles, guidelines or the rotation pivot.

Coordinate contract:

• position is the clip's anchor point (style.getPosition() semantics) and is the pivot that rotation (and any scale-from-anchor transform) happens around — NOT center. They coincide for all clip types with anchor (0.5, 0.5); do not rely on that.
• anchor is normalized (0..1): where position sits inside the box. (0.5, 0.5) for most clips, (0.5, 1.0) for subtitles (bottom-center).
• center is the geometry-box center, for layout, guidelines and snapping: center = [position[0] + (0.5 − anchor[0]) × width, position[1] + (0.5 − anchor[1]) × height].

Deriving the unrotated box: topLeft = [position[0] − anchor[0] × width, position[1] − anchor[1] × height] Corners are then rotated around position (the anchor point) — NOT around the box center — by rotation.

Deriving the visual (padding-inclusive) box when contentPadding is present: expand each side of the geometry box by its padding; that box's center is center + [(pad.right − pad.left) / 2, (pad.bottom − pad.top) / 2].

Freshness contract: the returned values are derived from the clip's logical style state, which style setters update synchronously — reading immediately after setScale/setWordWrapWidth/etc. reflects the change.