Wgsl GS chunk updates

[slimbuck]

This PR brings the wgsl GS chunks inline with glsl.

A separate PR will add the missing SOGS wgsl chunks.

[Copilot]

Pull Request Overview

This PR updates the WGSL GS chunks to align with the GLSL implementation, refactoring functions to return values rather than using output pointers and replacing the deprecated quatToMat3 implementations.

• Refactored fetch functions to return structured data (ScaleAndSH/SH) in gsplatSH.js
• Updated quaternion-to-matrix conversion usage in gsplatData.js and consolidated a new quatToMat3 in gsplatCommon.js
• Removed the old quatToMat3 in gsplatCompressedData.js

Reviewed Changes

Copilot reviewed 4 out of 4 changed files in this pull request and generated 1 comment.

File	Description
src/scene/shader-lib/chunks-wgsl/gsplat/vert/gsplatSH.js	Introduces new struct types and refactored fetch functions for spherical harmonic coefficients
src/scene/shader-lib/chunks-wgsl/gsplat/vert/gsplatData.js	Updates readCovariance to use a swizzled quaternion from unpackRotation
src/scene/shader-lib/chunks-wgsl/gsplat/vert/gsplatCompressedData.js	Removes the outdated quatToMat3 implementation
src/scene/shader-lib/chunks-wgsl/gsplat/vert/gsplatCommon.js	Adds a new implementation for quatToMat3 using WGSL’s new syntax

Comments suppressed due to low confidence (1)

src/scene/shader-lib/chunks-wgsl/gsplat/vert/gsplatData.js:24

• The addition of the .wxyz swizzle on the quaternion from unpackRotation changes the expected component order; please verify that this reordering is intended and consistent with the new quatToMat3 implementation.

let rot: mat3x3f = quatToMat3(unpackRotation(vec3f(unpack2x16float(bitcast<u32>(tAw)), tB.w)).wxyz);

[Copilot]

There are multiple implementations of quatToMat3 across the module; consider consolidating the quaternion-to-matrix conversion into a single shared implementation to avoid inconsistencies.