Source-Owned Skin Shell and Poster Composition

## Context Canonical Skin output currently starts at the controls root. Packaged video and live-video Skins instead own the player container boundary, projected media content, optional poster/placeholder behavior, and the inert overlay used by controls and gestures. React and HTML express this boundary differently: React accepts children, container props, poster render input, and placeholder styling; HTML projects default/poster slots through its Skin element. Canonical intent must preserve those public behaviors without exposing framework-specific markup. ## Acceptance criteria - [ ] Canonical source represents the Skin/container boundary without platform intrinsics. - [ ] React output preserves children, container props, class/style customization, poster input, and placeholder behavior. - [ ] HTML output preserves idiomatic media and poster projection without leaking generated internal identity. - [ ] Default, Minimal, and live video compositions can reuse the same shell where behavior is shared. - [ ] The visual overlay remains Skin-owned layout rather than a generic Core primitive. - [ ] React and HTML package/source-owned fixtures cover poster visibility and container composition. - [ ] Exact package, element, style, and registry closure is generated.