TSL: Improves compilation performance by 3.0x

[sunag]

Related issue: #31691

Description

Use webgpu_loader_materialx where it previously took on average ~1,550ms to compile all the materials in the scene on my i9 PC; currently, that same time is around ~515ms.

The main reason for the performance improvement was the caching of getNodeType(), where I achieved about a ~2.7x improvement. The rest were some other minor refinements. This should be more noticeable in code that involves many operations.

After some later revisions I introduced generateNodeType() into the class; it seems that the name describes well what happens, whereas the previous getNodeType is oriented toward requests where there is a cache layer.

• cache node types
• create uuid only when used
• use id as node hash

The tests were performed by summing all builds directly in NodeBuilder.build().

[github-actions]

📦 Bundle size

Full ESM build, minified and gzipped.

	Before	After	Diff
WebGL	359.27 85.31	359.27 85.31	+0 B +0 B
WebGPU	629.93 174.99	630.28 175	+347 B +7 B
WebGPU Nodes	628.51 174.74	628.86 174.75	+347 B +8 B

🌳 Bundle size after tree-shaking

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

	Before	After	Diff
WebGL	491.03 119.72	491.03 119.72	+0 B +0 B
WebGPU	703.55 189.99	703.88 190.04	+331 B +51 B
WebGPU Nodes	652.76 177.37	653.11 177.48	+352 B +110 B

[wyliefoxxx]

You deserve another round of applause 👏