Add Chinese design doc for mask and visual effects (NarraLeaf React)

[helloyork]

Motivation

• Provide a unified design and phased implementation plan for masks, clips, filters, blends, overlays and mask-based transitions to avoid multiple incompatible mask systems.
• Define goals to keep the implementation low-invasion, serializable, resource-preloadable and reusable across Displayable/Layer/Scene scopes.
• Propose a practical MVP and multi-phase rollout to enable immediate CSS-based capabilities while deferring higher-risk features like video alpha/canvas masks.

Description

• Add a new documentation file docs/mask-research.zh-CN.md that describes API proposals, design goals and implementation phases for visual effects and masking.
• Specify a VisualEffect/Transform field set (example VisualEffectTransformProps) and a unified DOM style resolver mapping mask/clip/filter/blend fields to maskImage, clipPath, filter, backdropFilter, mixBlendMode, etc.
• Recommend a chainable Displayable API (e.g. displayable.effect(), displayable.mask(), displayable.clip()) and an effect layer / OverlayEffect model for non-blocking video/textures, plus MaskTransition and Preset APIs.
• Outline a nine-phase rollout (Phase 1..9) and an MVP checklist that prioritizes CSS masks, clip-path, filter, mix-blend-mode, preload of mask images, an effect layer and a few presets like circleReveal and vignette.

Testing

• No automated tests were required or run for this documentation-only change.

---

Codex Task