TSL: Add bitcast functions and transpiler support

[cmhhelgeson]

Related issue: #31735

Description

Adds specialized blocks for translating bitcast functionality.

[sunag]

Instead of modifying the transpiler, could you create these functions as aliases in BitcastNode?

I think the GLSL functions have better names and this would also make migration easier since the transpiler is optional.

[github-actions]

📦 Bundle size

Full ESM build, minified and gzipped.

	Before	After	Diff
WebGL	338.91 79.11	338.91 79.11	+0 B +0 B
WebGPU	579.29 159.67	579.78 159.79	+490 B +120 B
WebGPU Nodes	577.9 159.43	578.39 159.55	+490 B +120 B

🌳 Bundle size after tree-shaking

Minimal build including a renderer, camera, empty scene, and dependencies.

	Before	After	Diff
WebGL	470.78 113.9	470.78 113.9	+0 B +0 B
WebGPU	649.44 175.54	649.44 175.54	+0 B +0 B
WebGPU Nodes	603.54 164.69	603.54 164.69	+0 B +0 B

[cmhhelgeson]

Example Code:

// ToVar() added for clarity
const instanceIndexBitcastFloat = uintBitsToFloat( float( instanceIndex ) );
const instanceIndexBitcastUint = floatBitsToUint( uint( instanceIndexBitcastFloat ) );
const floatToFloat = bitcast( instanceIndex, 'uint' ).toVar();

// vectors

const vec2Example = uintBitsToFloat( vec2(instanceIndex));
const vec2NativeBitcast = bitcast(instanceIndex, 'vec2')

// Note how with GLSL alias functions, inputType of original GLSL function is enforced
nodeVar0 = bitcast<f32>( u32( f32( instanceIndex ) ) );
nodeVar1 = bitcast<u32>( f32( u32( nodeVar0 ) ) );
// Whereas WGSL implementation is more freeform
nodeVar3 = bitcast<u32>( instanceIndex );


// vectors

nodeVar4 = bitcast<vec2<f32>>(vec2<u32>(instanceIndex, instanceIndex));
// regular bitcast will just translate what is given to it
nodeVar5 = bitcast<vec2<f32>>(instanceIndex);